Pharmacoproteomics of a Metalloproteinase Hydroxamate Inhibitor in ...

MOLECULAR AND CELLULAR BIOLOGY, Aug. 2008, p. 4896–4914 0270-7306/08/$08.00⫹0 doi:10.1128/MCB.01775-07 Copyright © 2008, American Society for Microbiology. All Rights Reserved.

Vol. 28, No. 15

Pharmacoproteomics of a Metalloproteinase Hydroxamate Inhibitor in Breast Cancer Cells: Dynamics of Membrane Type 1 Matrix Metalloproteinase-Mediated Membrane Protein Shedding䌤‡ Georgina S. Butler,1 Richard A. Dean,1 Eric M. Tam,2† and Christopher M. Overall1,2* Departments of Oral Biological and Medical Sciences1 and Biochemistry and Molecular Biology,2 Centre for Blood Research, Life Sciences Centre, University of British Columbia, Vancouver, Canada Received 27 September 2007/Returned for modification 3 November 2007/Accepted 18 May 2008

Broad-spectrum matrix metalloproteinase (MMP) inhibitors (MMPI) were unsuccessful in cancer clinical trials, partly due to side effects resulting from limited knowledge of the full repertoire of MMP substrates, termed the substrate degradome, and hence the in vivo functions of MMPs. To gain further insight into the degradome of MMP-14 (membrane type 1 MMP) an MMPI, prinomastat (drug code AG3340), was used to reduce proteolytic processing and ectodomain shedding in human MDA-MB-231 breast cancer cells transfected with MMP-14. We report a quantitative proteomic evaluation of the targets and effects of the inhibitor in this cell-based system. Proteins in cell-conditioned medium (the secretome) and membrane fractions with levels that were modulated by the MMPI were identified by isotope-coded affinity tag (ICAT) labeling and tandem mass spectrometry. Comparisons of the expression of MMP-14 with that of a vector control resulted in increased MMP-14/vector ICAT ratios for many proteins in conditioned medium, indicating MMP-14mediated ectodomain shedding. Following MMPI treatment, the MMPI/vehicle ICAT ratio was reversed, suggesting that MMP-14-mediated shedding of these proteins was blocked by the inhibitor. The reduction in shedding or the release of substrates from pericellular sites in the presence of the MMPI was frequently accompanied by the accumulation of the protein in the plasma membrane, as indicated by high MMPI/vehicle ICAT ratios. Considered together, this is a strong predictor of biologically relevant substrates cleaved in the cellular context that led to the identification of many undescribed MMP-14 substrates, 20 of which we validated biochemically, including DJ-1, galectin-1, Hsp90␣, pentraxin 3, progranulin, Cyr61, peptidyl-prolyl cis-trans isomerase A, and dickkopf-1. Other proteins with altered levels, such as Kunitz-type protease inhibitor 1 and beta-2-microglobulin, were not substrates in biochemical assays, suggesting an indirect affect of the MMPI, which might be important in drug development as biomarkers or, in preclinical phases, to predict systemic drug actions and adverse side effects. Hence, this approach describes the dynamic pattern of cell membrane ectodomain shedding and its perturbation upon metalloproteinase drug treatment. cleavage of native type I collagen by MMP-14 (128, 130). As their name suggests, the MMPs were originally thought to act predominantly upon extracellular matrix molecules. However, these proteases are now known to process a plethora of bioactive molecules (chemokines, cytokines, growth factors and their binding proteins, receptors, protease inhibitors, and cell adhesion molecules) (17, 75, 94), regulating processes involved in development (95) and normal cellular and extracellular homeostasis (88), as well as pathological processes such as inflammation and cancer (31, 41, 76, 91, 97, 122). MMPs are, therefore, promising drug targets for many diseases. Nonetheless, identification of the tissue-protective functions of MMPs has revealed important new roles in physiological function and homeostasis, the inhibition of which must be viewed with caution (31, 92). However, the full range of substrates, termed the substrate degradome (72), for members of this protease family is far from fully annotated. Hence, high-throughput nonbiased techniques to uncover novel substrates that might reveal new functional roles for MMP processing in physiological and pathological processes are required (90). Despite the fact that a large number of phase III clinical trials of small-molecule and peptidic MMP inhibitors (MMPI) were reached, these inhibitors were unsuccessful in treating cancer and arthritis, with some patients experiencing adverse

The membrane type 1 matrix metalloproteinase (MMP-14) is a member of the MMP family of zinc-dependent endoproteinases that are important in cancer (31, 92), consisting of 23 members in humans, 6 of which (including MMP-14) are membrane anchored (50, 122). The MMPs are one of four families of the metzincin clan, which includes ADAM (a disintegrinlike and metalloprotease domain) and ADAM with thrombospondin type 1 motifs (ADAM-TS) extracellular proteases (40). MMP-14 is constitutively activated in the Golgi apparatus by furin and, at the cell surface, undergoes autodegradation which removes and inactivates the catalytic domain, leaving a remnant ectodomain composed of the linker and hemopexin C domains (50). The hemopexin C domain encompasses binding sites known as exosites that localize substrates in the vicinity of the catalytic domain to improve catalysis (89), for example, * Corresponding author. Mailing address: Centre for Blood Research, Life Sciences Institute, Room 4.401, 2350 Health Sciences Mall, University of British Columbia, Vancouver, B.C. V6T 1Z3, Canada. Phone: (604) 822-2958. Fax: (604) 822-7742. E-mail: chris.overall @ubc.ca. † Present address: Dept. of Protein Engineering, Genetech Inc., South San Francisco, CA 94080. ‡ Supplemental material for this article may be found at http://mcb .asm.org/. 䌤 Published ahead of print on 27 May 2008. 4896

VOL. 28, 2008

PHARMACOPROTEOMICS OF A METALLOPROTEINASE INHIBITOR

side effects including tendonitis and myalgia (11, 22, 92, 93, 149). When the MMPI drug programs were initiated, it was rightly assumed that the role of MMPs in cancer was to enable tumor cells to degrade basement membrane and thus to metastasize. However, only three MMPs were recognized at the time, and other activities of MMPs were unknown and therefore not considered. Target validation against a few target and countertarget proteins (proteins that were related but with no strong role in the disease, nor ones which produced adverse effects [92]) in biochemical and cell culture assays provides useful but limited information, since these data fail to take into account that the target is embedded in a network of interactions and interconnected pathways within a complex proteome. It is the perturbation of such a web by the actions of a drug, as well as the blockade of antitargets (molecules related to the target that have protective functions in disease), that may lead to poor drug efficacy due to counterbalancing target inhibition and clinically unacceptable side effects or toxicity (92). As complex systems, animal models are important, yet they are limited by distinct differences between genotypes and proteomes, and life spans are short compared with those of humans, which impacts disease pathogenesis. For example, in rodents, there are more than 80 additional protease genes (102), a major class of enzyme drug targets (69, 92), many of which are expressed in host defense cells. Hence, there is a need for new system-wide drug target validation approaches, where both known and otherwise unpredictable targets can be monitored preclinically. Proteomics offers new high-content techniques for protease substrate discovery in complex cellular systems, as we have demonstrated previously using isotope-coded affinity tag (ICAT) labeling and tandem mass spectrometry (MS-MS) (25, 129) and isotope tags for relative and absolute quantitation (iTRAQ) labeling (26). Proteomics also has the potential to improve preclinical drug assessment to enhance the selection of promising drugs for the more-resource-consuming clinical studies. Here, we have utilized ICAT for assessing drug actions in cell culture. ICAT is used for quantitative comparison of the proteins in two samples labeled with a heavy or light label, for example, protease versus null or inactive protease (25, 129) or drug- versus vehicle-treated cells. Identical peptides from proteins originating from the two different samples are identified by the isotopic mass differences of the two labels, with the area of the two peaks allowing relative quantification of the peptides and, hence, the parent protein for the two samples. Peptides can then be sequenced by MS-MS to identify the parent protein (42). We were the first to use this technique to demonstrate an increase in the degradation of secreted protein substrates and proteolytic shedding of other substrates, many of which were novel, to the conditioned medium upon expression of MMP-14 in the human breast cancer cell line MDA-MB-231 compared with that in control cells transfected with vector or a catalytically inactive MMP-14 mutant (129). To investigate proteolysis in a cellular environment, protease inhibitors or RNA interference experiments can also be utilized to reduce the activity of a particular protease and thereby identify substrates which accumulate while providing information useful for drug development (14). Here, we have blocked protein shedding from MMP-14-transfected MDA-MB-231 cells, using an MMPI, and

4897

we have used proteomics to characterize the proteome-wide effects of the MMPI treatment, as well as to identify substrates, both known and novel. This analysis indicated unexpected and otherwise unpredictable drug actions. We also report that the combined analysis of ICAT ratios of proteins in cells expressing a protease, and before and after drug treatment, validates the substrate discovery. This proteomics approach describes a complex dynamic environment of basal proteolytic shedding from the plasma membrane and pericellular matrix by using metalloproteinase activity. In so doing, the precise processing of a diverse array of bioactive molecules that precisely alter their functions indicates key regulatory roles for MMPs in cell homeostasis and cancer. MATERIALS AND METHODS Proteins and peptides. Recombinant human MMP-14 (soluble form), MMP-2, MMP-1, and MMP-8 were expressed and purified as previously described (15, 16). Recombinant human MMP-9 was a gift from R. Fridman, Wayne State University, Detroit, MI. Recombinant human progranulin and rabbit polyclonal anti-progranulin antibody were gifts from A. Bateman, McGill University, Montreal, Canada. Niemann-Pick type C2 (NPC2) and DJ-1, expressed in Escherichia coli as a fusion protein C-terminal to glutathione S-transferase (GST) (Schistosoma japonicum) were gifts from P. Lobel, UMDNJ-R. W. Johnson Medical School, Piscataway, NJ, and M. Cookson, NIA, Bethesda, MD, respectively. Recombinant human elafin and anti-elafin antibody were gifts from G. Tremblay, Universite´ Laval, Quebec, Canada. Recombinant human dickkopf-1, follistatinlike 3, Kunitz-type protease inhibitor 1, iduronate-2-sulfatase, and pentraxin 3 were purchased from R&D Systems Inc. (MN). Human neuron-specific enolase was from USBiological (MA). Human beta-2-microglobulin and cysteine-rich protein 61 (Cyr61) were from Calbiochem (EMD Chemicals, CA) and PeproTech, Inc. (NJ), respectively. Human galectin-1 was from Research Diagnostics Inc. (NJ), and heat shock protein 90␣ (Hsp90␣) was from Stressgen Bioreagents Corp. (Victoria, BC, Canada). Peptidyl-prolyl cis-trans isomerase A (cyclophilin A) was from BIOMOL International (PA). Recombinant TSP-1 was from EMP Genetech (Ingolstadt, Germany). Peptides spanning the urokinase plasminogen activator receptor (uPAR) D1-D2 linker MMP cleavage site 81SGRAVTYSRS RYLEC95 and the D3 juxtamembrane plasmin cleavage site 272NHPDLDVQY RSG283 (7) were synthesized by Sigma Genosys (Oakville, ON, Canada). Antibodies. Anti-Axl receptor tyrosine kinase (catalog no. AF154) and anticysteine-rich motor neuron 1 (CRIM-1) (catalog no. AF1917) extracellular domain goat polyclonal antibodies were purchased from R&D Systems Inc. (MN). Anti-polyhistidine mouse monoclonal antibody was from Calbiochem (EMD Chemicals, CA). Rabbit polyclonal anti-cyclophilin A (catalog no. SA-296) was from BIOMOL International (PA). MAII, a mouse monoclonal antibody which recognizes the heparin-binding domain of TSP-1 was a gift from W. Frazier, Washington University School of Medicine, St. Louis, MO. Bric 229, a mouse monoclonal antibody raised against CD59 was from IBGRL (Bristol, UK). Horseradish peroxidase-conjugated and Alexa Fluor 680-conjugated secondary antibodies were from Bio-Rad Laboratories Ltd. (ON, Canada) and Molecular Probes (Invitrogen Canada Inc., ON, Canada), respectively. Cell culture. MDA-MB-231, a human breast cancer cell line stably transfected with human MMP-14 (with a FLAG tag in the juxtamembrane stalk) or the empty vector, cells were cultured in Dulbecco’s modified Eagle medium (DMEM)–10% fetal bovine serum with G418 selection (geneticin, 1 mg/ml) (Gibco BRL, Invitrogen Canada Inc., ON, Canada), as described previously (130). Drug treatment and preparation of conditioned media. Semiconfluent cells in roller bottles (850 cm2; Becton Dickinson ON, Canada) or T175 flasks were washed three times with phosphate-buffered saline (PBS) and incubated overnight in phenol red-free, serum-free DMEM (Gibco BRL, Invitrogen Canada Inc., ON, Canada) to remove bovine serum proteins, thus increasing the specific labeling of cellular proteins. Following further PBS washes, cells were incubated in 50 ml of phenol red-free DMEM containing the nonpeptidic hydroxamate MMPI prinomastat {drug code, AG3340; (3S)-N-hydroxy-4-(4-(pyrid-4-yloxy)benzenesulfonyl)-2,2-dimethyl-tetrahydro-2H-1,4-thiazine-3-carboxamide [116, 117], 10 ␮M final concentration, dissolved in dimethyl sulfoxide [DMSO], 0.1% [vol/vol] final concentration, or DMSO vehicle, 0.1% [vol/vol] final concentration} alone. Medium was harvested after 48 h, and protease inhibitors (10 ␮M EDTA, 1 mM phenylmethylsulfonyl fluoride, 10 ␮M leupeptin, 10 ␮M pepstatin

4898

BUTLER ET AL.

A [final concentrations]) were added. The conditioned medium was clarified by centrifugation for 20 min at 500 ⫻ g at 4°C and filtration (0.22-␮m filter). The medium was then concentrated 100- to 400-fold using a 3-kDa cutoff membrane (Centriprep and Microcon; Amicon). Protein was quantified by using a microBCA assay (Pierce Biotechnology Inc., IL), and samples were frozen at ⫺70°C prior to ICAT labeling. Preparation of membrane-enriched fraction. Following the collection of conditioned medium, cells were washed three times with PBS and detached in versine (140 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 䡠 7H2O, 1.8 mM NaH2PO4, 0.5 mM EDTA, 1.1 mM glucose [pH 7.4]) with an additional 5 mM EDTA. Harvested cells were washed with cold membrane buffer (50 mM Tris [pH 6.8], 200 mM NaCl, 10 mM CaCl2) and then resuspended in membrane buffer containing protease inhibitors as described above. To avoid introducing detergents into the proteome preparations, cells were lysed by nitrogen decompression. Cells on ice were pressurized to 500 lb/in2 using N2 in a cell disruption bomb (Parr Instrument Co., Moline, IL) for 30 min. Following rapid decompression and cell lysis, the cell lysate was centrifuged at 1,100 ⫻ g for 10 min at 4°C. The supernatant was centrifuged at 48,000 ⫻ g for 60 min at 4°C, and the membrane-enriched pellet was resuspended in freshly prepared buffer (200 mM Tris [pH 8.8], 5 mM EDTA, 6 M urea, 0.05% sodium dodecyl sulfate [SDS]) and homogenized (11,500 rpm, 5 s; Polytron unit). Protein was quantified and frozen as above. Cell viability and apoptosis assays. MDA-MB-231 cells (transfected with MMP-14 or an empty vector) were treated with the hydroxamate inhibitor prinomastat (10 ␮M final concentration), with vehicle (0.1% [vol/vol] DMSO), or with 1,10-phenanthroline (0.1 mM or 0.5 mM), or cells were left untreated in serum-free medium for 48 h. Cells were assessed for morphology, as visualized by light microscopy; for cellular proliferation, using a 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide (MTT) assay (ATCC, VA), which measures mitochondrial reductase function, or a Cyquant assay (Molecular Probes, Invitrogen Canada Inc., ON, Canada), which quantifies nucleic acids; and for apoptosis, by measuring caspase 3 activity with a fluorogenic caspase 3 substrate (N-acetyl-Asp-Glu-Val-Asp-AMC [7-amino-4-methylcoumarin]; BD Biosciences, ON, Canada). Specificity was confirmed by using a caspase inhibitory peptide (N-acetyl-Asp-Glu-Val-Asp-CHO; BD Biosciences, ON, Canada). All assays were carried out by following manufacturers’ instructions. ICAT labeling and MS. Proteins (100 ␮g) from MMPI-treated or vehicletreated MDA-MB-231 cells (from conditioned medium or membrane fractions) were labeled at the cysteine residues with isotopically heavy 13C9- or light 13C0labeled cleavable ICAT reagent (Applied Biosystems Inc., CA) according to the manufacturer’s protocol. Three biological replicates for the conditioned media secretome and two for membrane fractions were labeled and analyzed as previously described (25, 129). Briefly, samples were separately denatured, reduced, and labeled with either the [13C9]ICAT or the [13C0]ICAT reagent. Then, samples were combined, digested with trypsin, and purified by strong cation-exchange chromatography. Labeled peptides were isolated via the tag’s biotin moiety by avidin affinity chromatography, and the ICAT labels were removed by acid treatment. Peptides were then fractionated by multidimensional liquid chromatography (LC) using strong cation-exchange and C18 columns (13). A technical replicate was performed with conditioned medium from experiment 1, where the strong cation-exchange elution conditions were altered from (i) a 0 to 500 mM gradient of ammonium acetate (pH 3) with 5% acetonitrile over 75 min at a flow rate of 6 ␮l/min to (ii) a 0 to 250 mM gradient of ammonium acetate (pH 3) with 5% acetonitrile over 75 min at a flow rate of 5 ␮l/min. Peptides were analyzed by nanospray MS, using a QStar Pulsar unit for quadrupole time-offlight (TOF) MS (Applied Biosystems Inc). MS-MS fragmentation (2 s; 65 to 1,800 m/z) was performed with four of the most intense ions, as determined from a 1-s survey scan (300 to 1,500 m/z). Peptide quantitation and bioinformatics. ICAT ratios between isotopically heavy and light tryptic peptides were calculated using ProICAT software (Applied Biosystems Inc.). Proteins were identified using Mascot software (Matrix Science, MA), querying the peptide sequences against those of the National Centre for Biotechnology Information nonredundant protein database (http: //www.ncbi.nlm.nih.gov/). All peptide identifications were confirmed manually using the Swiss Institute of Bioinformatics BLAST network service (http://us .expasy.org/). Peptides resulting from missed tryptic cleavages are included, as the samples are combined prior to tryptic digestion, and therefore, there is an equal probability that a cleavage will be missed in each sample. The consistency of the ratios between peptides from cleaved sites and those from missed sites demonstrates the validity of including these peptides, and analyzing these as separate peptides increases confidence in the identification of these proteins (see Tables S1 and S2 in the supplemental material).

MOL. CELL. BIOL. In vitro substrate validation. Candidate substrates were incubated for 18 h at 37°C in assay buffer (100 mM Tris-HCl, 30 mM CaCl2) in the presence of 1 mM amino-phenylmercuric acetate (APMA) to activate zymogens (pro-MMPs), with or without soluble MMP-14 (or other MMPs, as detailed in the text) at enzyme/ substrate molar ratios ranging from 1:1,000 up to 1:10 (as stated in the legends to Fig. 3 and 6). Digests were stopped by the addition of reducing SDS-polyacrylamide gel electrophoresis (PAGE) sample buffer and were analyzed by Trisglycine or Tris-Tricine SDS-PAGE and silver stained or Western blotted with appropriate antibodies. For some assays, the digests were mixed with the appropriate matrix (alpha-cyano-4-hydroxycinnamic acid for proteins and peptides of ⬍10 kDa or sinapic acid for those of ⬎10 kDa) and spotted onto matrix-assisted laser desorption ionization (MALDI) plates for analysis by MS. Masses of cleavage products were determined by MALDI-TOF MS on a Voyager-DE STR biospectrometry workstation (Applied Biosystems Inc., CA). Cleavage of uPAR peptides. Peptides (100 ␮M) representing sequences from the D3 juxtamembrane region of uPAR, 272NHPDLDVQYRSG283 (theoretical mass, 1,400.49 Da), or encompassing the known MMP cleavage site, D1-D2 linker 81SGRAVTYSRSRYLEC95 (theoretical mass, 1,747.97 Da), were incubated at 37°C in assay buffer alone or with 1 ␮M plasmin, chymotrypsin, or MMPs (with the addition of 1 mM APMA to activate pro-MMPs). After an 18-h incubation, the reaction mixtures and controls (equivalent to 50 pmol peptide) were mixed with alpha-cyano-4-hydroxycinnamic acid matrix and analyzed by MALDI-TOF MS.

RESULTS Cellular response to MMPI treatment. To gain further insight into the substrate degradome of MMP-14, we utilized the MMPI prinomastat to block metalloproteinase protein processing and ectodomain shedding in MDA-MB-231 breast cancer cells transfected with MMP-14. We used the drug vehicle DMSO as a control at an equivalent final concentration of 0.1% (vol/vol). A largely unresolved concern of the unsuccessful MMPI clinical trials has been whether the MMP drug targets were effectively dosed, i.e., were inhibitor concentrations systemically attained and available for a sufficient time (22)? Since MMP-14 autodegradation is blocked by MMPIs (50), we assessed the status of MDA-MB-231 cell surface MMP-14 after cells were treated for 48 h with 10 ␮M prinomastat or vehicle. Mature (⌬propeptide) MMP-14 accumulated in the membrane fraction in the presence of the MMPI, consistent with reduced autodegradation of the enzyme (see Fig. S1A in the supplemental material). Hence, at this drug dosage, MMP-14 and likely any other MMPs present have sufficient drug exposure to block activity. Consistent with extensive activity and pharmacokinetic toxicity studies that permitted the use of this and other hydroxamate MMPIs in clinical trials (22, 117), we confirmed that prinomastat at a 10 ␮M concentration had no effect on cell viability (see Fig. S1B to D in the supplemental material). No effects were apparent for appearance (see Fig. S1B in the supplemental material), proliferation (see Fig. S1C in the supplemental material), or apoptosis (see Fig. S1D in the supplemental material) in this system with prinomastat treatment, whereas treatment with 1,10-phenanthroline, a zinc chelator and nonspecific inhibitor of metalloproteinases, was toxic at concentrations of ⱖ0.1 mM. Thus, the changes in ICAT ratios between MMPI and vehicle were due to metalloproteinase inhibitory effects and not due to drug-induced alterations in cell proliferation, apoptosis, or toxicity to the MDA-MB-231 cells. ICAT analysis and protein identifications. Three biological replicate experiments were performed in which proteomes

VOL. 28, 2008

PHARMACOPROTEOMICS OF A METALLOPROTEINASE INHIBITOR

4899

TABLE 1. Number of proteins identified in each ICAT experimenta Proteins identified

Total MMPI/vehicle ICAT ratio ⱕ0.77 MMPI/vehicle ICAT ratio ⱖ1.50

ⱖ99 ⱖ95 ⱖ99 ⱖ95 ⱖ99 ⱖ95

No. of proteins in membrane preparations

No. of proteins in conditioned medium

Confidence level (%) Expt 1a

Expt 1b

Expt 2

Expt 3

Expt 1

Expt 2

161 190 108 123 1 2

188 214 122 137 3 3

108 124 43 53 16 19

117 139 96 112 2 2

158 194 37 46 22 25

224 265 47 59 41 50

a Shown is the number of proteins identified from ICAT-labeled tryptic peptides from conditioned medium or membrane preparations of MMP-14-transfected MDA-MB-231 breast cancer cells in each experiment at confidence levels assigned by Mascot of ⱖ99% and ⱖ95%. Full lists of proteins and peptides are given in Tables S1 and S2 in the supplemental material. The ICAT ratio cutoff for decreased ICAT ratios of ⱕ0.77 was based upon ratios obtained for known substrates (see text and Table 2 for details). For high ratios, a cutoff of ⱖ1.5 was chosen, as previously defined (129). Experiments 1a and 1b are technical replicates in which the same ICAT-labeled conditioned medium sample was subjected to different strong cation-exchange separation (1a, an elution gradient of 0 to 500 mM salt; 1b, an elution gradient of 0 to 250 mM salt) prior to mass spectrometry (see Table S3 in the supplemental material).

from MMPI or vehicle-treated MMP-14 cell transfectants were compared. The number of proteins identified in conditioned medium and cell membrane fractions in each of the three experiments at 48 h is shown in Table 1. Complete lists of the peptides and proteins identified for each experiment are presented in Tables S1 (conditioned medium) and S2 (membrane fractions) in the supplemental material. For conditioned medium, a total of 519 individual peptides were identified in the three experiments, which encompassed 269 different proteins; 65% of these proteins were identified in only one of the experiments, 28% were identified in two experiments, and 7% were identified in all three experiments, demonstrating the need for biological replicates to maximize proteome coverage. We also performed technical replicates in which duplicate separations by strong cation exchange of conditioned medium sample 1 were performed that differed only in the length of the elution gradient. In this technical replicate, around 400 peptides were identified in each of the analyses: 33% of the peptides were common to the two analyses, and these had highly consistent ICAT ratios (see Table S3 in the supplemental material); 35% and 32% of peptides were unique to the 0 to 500 mM salt gradient (Table 1, conditioned medium sample 1a) and 0 to 250 mM salt gradient (Table 1, conditioned medium sample 1b), respectively. Differences in the peptides identified, even within the same biological sample, result from undersampling during mass spectrometry, i.e., a technical inability, partly due to MS duty cycle limits, to identify every peptide in a complex sample (63). Effects of MMPI on protein levels. When the secretome of MMP-14-transfected MDA-MB-231 cells was compared with that of a vector-transfected control, the shedding of MMP-14 substrates to the medium was increased, and soluble substrates that were degraded decreased in the medium (25, 129). We hypothesized that levels of MMP substrates and binding partners would change in MDA-MB-231 cell transfectants in the presence of the MMPI due to reduced processing, shedding, and clearance (Fig. 1). Soluble proteins in the secretome degraded by MMP-14 (ICAT ratio of MMP-14/vector, ⬍1) should undergo less degradation and clearance in the presence of the MMPI, so the ICAT ratio of MMPI/vehicle would be ⬎1 in the medium (Fig. 1A). Shedding of proteins to the conditioned medium (MMP-14/vector, ⬎1) would be inhibited by the MMPI, so there would be reduced cleavage and release of

membrane-associated proteins (Fig. 1B) or pericellular proteins (Fig. 1C and D) by MMPs. Hence, there would be less ectodomain in the medium (MMPI/vehicle, ⬍1) and, correspondingly, more intact substrate associated with the cell membrane (MMPI/vehicle, ⬎1). Levels of proteins that are bound to a shed protein but are not themselves substrates would also be modulated in this way, for example, ligands of cleaved receptors (Fig. 1B). A similar ratio might also result from dominant-negative effects of the inhibited MMP-14 (Fig. 1E): an MMPI blockade of MMP-14 autodegradation (50, 133) would result in the accumulation of mature inhibited MMP-14 on the cell surface. This might titrate substrates and interacting proteins which localize to MMP-14 exosites from the conditioned medium (MMPI/vehicle, ⬍1) to the membrane (MMPI/ vehicle, ⬎1). Of the proteins identified at ⱖ99% confidence, a 58% identification was based on one peptide, as is typical for ICAT experiments, 24% was based on two peptides, and 18% was based on more than two peptides (averages for the three conditioned medium biological replicates and two technical replicate experiments). Most proteins do not contain large numbers of cysteine residues (35% contain only one, and 9% have none, including type I collagen, an important MMP-14 substrate). Hence, although three biological replicates were performed for conditioned medium, the detection of only one or two ICATlabeled tryptic peptides for 82% of the proteins precluded the calculation of standard deviations for the heavy/light-label ratios. Instead of setting rigid numerical (and rather arbitrary) limits to decide whether a ratio was significantly altered or not, and thus whether the protein in question was likely to be a substrate of MMP-14, the ICAT ratio “cutoff” was based around ICAT ratios measured for known substrates identified in this cellular system. Metalloprotease substrate identification. To validate the analysis of MMPI effects as a pharmacoproteomic confirmation of a degradomic substrate screen, we looked for evidence of reduced processing, shedding, and clearance of known MMP substrates. Twenty-nine known MMP substrates were identified. These proteins had average MMPI/vehicle ICAT ratios of ⱕ0.77 (Table 2; also Table S4 in the supplemental material, which list sequences and ratios of every peptide identified in this study), indicating a decreased concentration of these proteins in the conditioned medium in the presence of

4900

BUTLER ET AL.

MOL. CELL. BIOL.

FIG. 1. Hypothesis: the MMPI attenuates shedding and release of cleaved proteins into the conditioned medium. Without MMP-14 (left panels, ⫺MMPI ⫹ vector), no MMP-14-mediated processing occurs. With MMP-14 but in the absence of the MMPI (center panels, ⫺MMPI ⫹MMP-14), active MMP-14 on the cell membrane (A) processes secreted proteins, which may result in further cleavages and clearance by MMPs or other proteases; (B) sheds membrane-associated or integral membrane proteins or their binding partners from the cell surface; (C) processes or releases proteins from extracellular and pericellular matrix; or (D) sheds directly or indirectly mobilizes secreted proteins from cell binding sites, e.g., by processing proteoglycans or integrins. These events will be blocked by a broad-spectrum MMPI (right panels, ⫹MMPI ⫹MMP-14). In the presence of an MMPI, soluble substrates increase in the conditioned medium (A). Whether the ratio changes or not will depend upon the rate of clearance of any fragments which will still be quantified as labeled tryptic peptides. Previously shed cell- or matrix-associated proteins decrease in the conditioned medium (B, C, and D), which coincides with their increase in the membrane or matrix. A similar response might be caused by MMPI-induced dominant-negative effects (E). Autodegradation of MMP-14 (center panel) is prevented by the MMPI, leading to an accumulation of mature MMP-14 at the cell surface (right panel). These inhibited MMP-14 molecules could act as “substrate traps,” binding substrates (and other interacting molecules) at exosites without cleavage and release. Hence, shed and soluble proteins would be titrated from the conditioned medium and sequestered at the cell surface. The predicted ICAT ratios for cells transfected with MMP-14 compared with empty vector (MMP-14/vector) and cells transfected with MMP-14 treated with inhibitor drug or vehicle (MMPI/vehicle) are shown adjacent to each panel for proteins in the conditioned medium (Medium) or cell membrane fractions (Membrane).

the inhibitor. This suggests a reduction in shedding from pericellular sites (cell membrane and pericellular matrix) or binding to the inhibited form of MMP-14 that would titrate proteins from the medium without cleavage (Fig. 1). Eleven of these established MMP substrates are known to be processed by MMP-14 (Table 2, references). For the other 18, cleavage by MMP-14 has not been reported, but based on the redundancy of processing by the MMP family, it is likely that many of these are MMP-14 substrates. Indeed, biochemical analyses of two of these proteins, galectin-1 and Hsp90␣, revealed that they are also substrates of MMP-14 (Fig. 2). Galectin-1, a lectin involved in the regulation of cell adhesion, migration, and proliferation (103), was processed by MMP-14 in a concentration-dependent manner from an apparent mo-

lecular mass of 11.5 kDa to 8.9 kDa. Hsp90␣, a cytoplasmic molecular chaperone and extracellular regulator of cell invasion (34), was processed from an apparent molecular mass of 96.6 kDa to a fragment of 79.8 kDa. Follistatin-related protein 1, cystatin C, and GRO␣, however, were not processed by MMP-14 in vitro (data not shown), suggesting that these ICAT ratios were lowered due to indirect effects of the MMPI, inhibition of other active metalloproteases expressed by these cells, or binding to MMP-14 exosites or suggesting that essential proteins or interactions present in the cellular context are not reproduced in the biochemical assays. Specificity of prinomastat for metalloproteinases. The quest for MMPIs has yet to reveal any which are 100% specific for an individual MMP (22, 92, 93). Prinomastat potently inhibits MMP-

VOL. 28, 2008

PHARMACOPROTEOMICS OF A METALLOPROTEINASE INHIBITOR

4901

TABLE 2. Known substrates of MMPs identified in conditioned medium of MMP-14-transfected MDA-MB-231 cells treated with MMPIa MMPI/vehicle ratio

No. of peptides

Known MMP-14 substrates CTGF Laminin ␥2 chain Complement C3 Amyloid ␤ A4 Fibronectin Fibrillin-1 KiSS-1 C1qBP Fibrinogen ␤ chain MMP-14 Interleukin-8

0.22 0.32 0.36 0.37 0.50 0.60 0.65d 0.71 0.71 0.75 0.77

12 2 7 5b 4 4 1 1 1 2 1

Known MMP substrates Follistatin-related protein 1 VEGF-A Latent TGF ␤-binding protein 1 Tissue factor pathway inhibitor Galectin-1 GRO␣ uPA IGFBP-1 Agrin IGFBP-3 Cystatin C Fibromodulin uPAR Hsp90␣ Collagen ␣1 (VI) chainc MMP-1 Endothelin-1 Dystroglycan

0.33 0.37 0.38 0.40 0.42 0.43 0.43 0.43 0.47 0.56 0.56 0.58 0.58 0.61 0.62 0.71 0.74 0.75

8 2 3 2 4 5 3 2 3 1 1 2b 1 6b 1 2 2 1

Protein

MMP(s) responsible

1, 3, 2, 3, 14 1, 2, 1, 2, 2, 3, 2, 9, 14 3, 7, 14 8, 9,

7, 13, 14 12, 13, 14, 20 3, 14 3, 7, 14, 15, 19, 25 9, 12, 13, 14 14, 16, 24 8, 12, 13, 14, 17, 25 14

2 3, 7, 9, 19, 1, 16 2, 9 1, 7–9, 12 2 9 3 3, 9 3 1, 2, 3, 7, 9, 19 2 13 3, 12, 19, 25 2 2 3, 7, 10 2 2, 9

Reference(s)

43, 129 39, 61, 62, 101 108, 129 51, 80, 123 29, 33, 37, 124 5 127 107 9, 10, 32, 33, 52 50, 133 129, 131, 137 25 64 24 8, 23 26 137 135 21, 62, 74 120, 138 36, 73, 79, 111, 142 25 46 4, 60 26 81 113, 121 35 143, 148

a MDA-MB-231 breast cancer cells transfected with MMP-14 were treated with the MMPI prinomastat or with DMSO vehicle under serum-free conditions for 48 h. Proteins in the conditioned medium samples were labeled with ICAT reagents and combined, and the biotin-tagged tryptic peptides were separated and analyzed by multidimensional LC–MS-MS. Peptides were identified at a ⱖ99% confidence level, unless otherwise noted in Table S1 and S4 in the supplemental material, which Table 2 summarizes. Values shown are the averages of ratios for peptides from three separate experiments (MMPI/vehicle). The individual peptide sequences are shown in Table S4 in the supplemental material. The number of unique peptides identified for each protein is shown. Abbreviations: C1qBP, complement component 1 Q subcomponent-binding protein; VEGF-A, vascular endothelial growth factor A; TGF, transforming growth factor; GRO␣, growth-related protein alpha; IGFBP, insulin-like growth factor-binding protein; Hsp, heat shock protein. b The number of peptides includes those differing only by oxidation of a methionine, which were counted as two peptides, since they were identified independently in the MS analysis. c The collagen ␣3 (type VI) chain is cleaved; therefore, the collagen ␣1 (type VI) chain is also likely cleaved. d The ratios obtained in two experiments were inconsistent, but the ratio from experiment 1 was chosen as it was derived from technical replicates which were reproducible (see Table S3 in the supplemental material).

FIG. 2. Confirmation of MMP-14 processing of galectin-1 and Hsp90␣ in vitro. (A) MMP-14 cleavage of galectin-1 with increasing enzyme/substrate molar ratio (1:500 to 1:10) was analyzed on 15% Tris-Tricine SDS-PAGE. (B) Analysis of MMP-14 proteolysis of Hsp90␣ on 10% SDS-PAGE. Arrows indicate cleaved protein fragments, and apparent molecular masses are shown. Positions of molecular mass markers (103 Da) are indicated.

14, as well as the gelatinases MMP-2 and MMP-9 and MMP-3 and MMP-13 (Ki values, 30 to 300 pM), but weakly inhibits MMP-1 and MMP-7 (see Fig. S2A in the supplemental material) (116, 117). Using quantitative reverse transcription-PCR (RTPCR) (83) and microarray analysis (C. J. Morrison and C. M. Overall, unpublished data), we found that MDA-MB-231 cells expressed several MMPs at the RNA level, although no soluble MMP activity was detected previously by using a peptide substrate-based assay (129). We detected peptides for MMP-1, MMP-3, and MMP-14, as well as for ADAM-10, in the conditioned medium (see Table S1 in the supplemental material) and MMP-1 and a sequence that was 77% identical to that of ADAM-29 in the membrane preparations (see Table S2 in the supplemental material). The MMP-1 peptide (CGVPDVAQFV LTEGNPR) spans both the Cys switch, responsible for enzyme latency, and the activation cleavage site (underlined). The homologous peptide from the MMP-3 propeptide was also detected (CGVPDVGHFR), confirming that these peptides were from the

4902

BUTLER ET AL.

zymogen form of the enzymes. The ADAM-10 peptide (YGPQ GGCADHSVFER) contains the cysteine switch of the propeptide, indicating that this enzyme is also present as a zymogen. No MMP-2 (also known as gelatinase A) peptides were detected. Although MMP-14 can activate MMP-2 (50), we have previously shown that MDA-MB-231 cells express a negligible amount of MMP-2 (129), and here we have confirmed this by zymography in the presence and absence of the MMPI (see Fig. S2B in the supplemental material). Hence, a few MMPs were identified as inactive proenzyme forms, whereas other active MMPs were either absent or present at low levels. However, the possibility that MMPs or metalloproteinases, such as ADAM-10, that are inhibited by prinomastat could contribute to substrate cleavage in vivo cannot be discounted. The relative importance of MMP-14 compared with that of other metzincins in vivo will depend upon the microenvironment and temporal-spatial expression pattern of the enzymes and each substrate. Validation of MMP-14 substrate shedding by using an MMPI. To enhance the identification of proteins that are shed from the cell membrane and pericellular environment by MMP-14 specifically, proteins in conditioned medium from MDA-MB-231 cells transfected with MMP-14 were compared with those from cells transfected with empty vector but in the absence of MMPI (Fig. 1 indicates predicted ICAT ratios, and see Table S5 in the supplemental material for a full list of the proteins and peptides identified). As we previously described (129), there were a number of proteins with ICAT ratios that were increased in the conditioned medium of MMP-14-transfected cells compared with those of the vector-transfected cells, indicating shedding by MMP-14 (Table 3 shows averaged ICAT ratios for each protein MMP-14/vector ratio; for the identities and ratios of their individual peptides, see Table S6A in the supplemental material). Significantly, many of these proteins, including seven that are known MMP substrates, had reduced ICAT ratios when the MMP-14-transfected cells were incubated with the MMPI compared with those of the vehicle (Table 3; also see Table S6A in the supplemental material, MMPI/vehicle). This trend (since these were separate experiments, the absolute values are not directly comparable) was apparent from comparing either the averages of ICAT ratios for all the peptides obtained for each protein (Table 3) or only the averages of ICAT ratios of the peptides that were common to the two analyses (see Table S6B in the supplemental material). Hence, the addition of MMPI to the MMP-14-transfected MDA-MB-231 cells blocked release or shedding of these proteins to the conditioned medium. This reversal of the ICAT ratios following the addition of a protease inhibitor to MMP-14-transfected cells is a strong validation that the high protease/vector ICAT ratios represent MMP-14 substrate cleavage and shedding. Novel proteins shed by MMP-14. The ICAT ratios for CRIM-1, a type I membrane protein which binds bone morphogenetic proteins (140), were confirmed by Western blotting (Fig. 3A). Stable expression of MMP-14 in the MDA-MB-231 cells resulted in an increase in levels of the 89-kDa CRIM-1 ectodomain (and also a smaller 51-kDa band, Fig. 3A, arrow) in the conditioned medium compared with those of vectortransfected cells (ICAT ratio MMP-14/vector, 1.51), suggesting a MMP-14-dependent increase in shedding. Levels of shed CRIM-1 ectodomain in the conditioned medium of MMP-14-

MOL. CELL. BIOL. TABLE 3. MMPI blocks shedding or release of proteins to the conditioned mediuma MMP-14/vector Protein

Known MMP substrates Fibromodulin Fibronectin MMP-14 MMP-1 CTGF Tissue factor pathway inhibitor Follistatin-related protein 1 Other bioactive molecules EGF-containing fibulin-like extracellular matrix protein 1 RNase (pancreatic) Quiescin Q6 Elafin RNase T2 CD59 Galectin-3-binding protein Ectonucleotide pyrophosphatase/ phosphodiesterase 1 IGFBP-7 Cysteine-rich motor neuron-1 Niemann-Pick disease, type C2 variant Hypothetical protein LOC196463 Iduronate 2-sulfatase TIMP-1 Serine protease 23 Pentraxin-related protein PTX3 N-Acetylglucosamine-6-sulfatase Follistatin-related protein 3 KIAA1392/Storkhead-box 2 Kunitz-type protease inhibitor 1

MMPI/vehicle

Ratio

No. of peptides

Ratio

No. of peptides

4.22 2.85 2.61 1.85 1.57 1.23 1.10

1 9 2 1 5 2 8

0.58 0.50 0.75 0.71 0.22 0.40 0.33

2b 4 2 2 12 2 7

3.90

2

0.25

1

3.05 2.16 1.85 1.71 1.67 1.61 1.58

2 1 2 2 1 2 1

0.22 0.72 0.53 0.71 0.41 0.51 0.54

1 5 2 1 2 1 2

1.54 1.51c 1.42

7 4 2

0.26 0.24 0.36

2 5 4b

1.33

1

0.57

1

1.32 1.32 1.32 1.26 1.19 1.14 1.14 1.13

1 3 2 3 1 1 1 3

0.09 0.61 0.36 0.51 0.72 0.41 0.39 0.31

1 8b 1 1 1 1 1 1

a A comparison of MDA-MB-231 cells transfected with MMP-14 to those transfected with empty vector (in the absence of inhibitor) (MMP-14/vector) revealed several proteins which were increased in the medium of MMP-14transfected cells, indicating increased shedding/release from cellular or pericellular sites that is MMP-14 dependent. A comparison of MMP-14-transfected MDA-MB-231 cells treated with MMPI or with DMSO vehicle (MMPI/vehicle) revealed proteins which were decreased in the conditioned medium, suggesting inhibition of metalloprotease-dependent shedding. The individual peptide sequences for MMP-14/vector are shown in Table S5 in the supplemental material, and those for MMP-14/vehicle are shown in Table S6 in the supplemental material. Abbreviations: CTGF, connective tissue growth factor; IGFBP, insulinlike growth factor binding protein; EGF, epidermal growth factor. b Peptide numbers include those differing only by oxidation of a methionine, which were counted as two peptides, since these are identified independently of each other in the MS analysis. c Peptide mapping (26) of the three peptides for this protein indicate shedding of the N-terminal domain. The most N-terminal peptide had a ratio of 3.06 compared with ratios of 1.08 and 0.40 for peptides nearer the C terminus and plasma membrane.

transfected cells were reduced by treatment with the MMPI (ICAT ratio MMPI/vehicle, 0.24), confirming that shedding had been inhibited. We chose several proteins for in vitro validation by cleavage assays to determine whether the shedding was direct, that is, due to MMP-14 cleavage of the proteins themselves or due to indirect effects or other proteases. Follistatin-related protein 3, which binds and masks the activities of activin, myostatin, and some bone morphogenetic proteins (119), was processed by

VOL. 28, 2008

PHARMACOPROTEOMICS OF A METALLOPROTEINASE INHIBITOR

4903

FIG. 3. Biochemical validation of novel MMP-14 substrates. (A) Conditioned medium proteins (20 ␮g) from MDA-MB-231 cells transfected with MMP-14 or empty vector (left panel, MMP-14/vector) and MMP-14-transfected cells treated with prinomastat (10 ␮M) or vehicle (right panel, MMPI/vehicle) were subjected to ICAT multidimensional LC–MS-MS analysis (ratios shown) or were separated by 11% SDS-PAGE. CRIM-1 was detected by Western blotting using an anti-CRIM-1 goat polyclonal antibody raised against the extracellular domain. (B) MMP-14 cleavage of recombinant follistatin-related protein 3 with a C-terminal His tag incubated with increasing enzyme/substrate molar ratios (1:100, 1:50, 1:10, and 1:5) was analyzed with 12.5% Tris-Tricine SDS-PAGE and silver stained or Western blotted (1:100, 1:50, and 1:10) with an anti-polyhistidine antibody. Comparison of these two analyses reveals near comigration of MMP-14 autolytic degradation products just above the 30.9-kDa band in the silver-stained gel. (C) Recombinant pentraxin 3 (Ptx3) with a C-terminal His tag was electrophoresed on 12.5% Tris-Tricine SDS-PAGE and subjected to Western blotting with an anti-polyhistidine antibody. (D) Recombinant human Niemann-Pick type C2 (NPC2) incubated with or without MMP-14 was electrophoresed on 12.5% Tris-Tricine SDS-polyacrylamide gels and silver stained. (E) Iduronate-2-sulfatase (IDS) samples, after incubation with MMP-14, were electrophoresed on 15% Tris-Tricine SDS-polyacrylamide gels and silver stained. All samples were incubated for 18 h at 37°C. Western blotting was carried out using appropriate Alexa-Fluor 680-conjugated immunoglobulin G secondary antibodies (Molecular Probes) and detected with an Odyssey infrared scanner (LiCor). MMP-14 did not cross-react with these antibodies. Arrows indicate cleaved protein fragments, and apparent molecular masses are shown. Positions of molecular mass markers as 103 Da are indicated.

MMP-14 in a concentration-dependent manner from 46.8-kDa to a 30.9-kDa C-terminal product, visualized by silver staining and Western blotting (Fig. 3B). Processing of follistatin-related protein 3 by MMP-2 was similar, but MMP-1 and -8 did not cleave the protein (see Fig. S3A and B in the supplemental material). Pentraxin 3, a multifunctional protein involved in innate immunity and inflammation (38), was processed by MMP-14 from an apparent molecular mass of 51.3 kDa to C-terminally His-tagged products of 34.3 kDa and 30.9 kDa, as

detected by Western blotting with an anti-polyhistidine antibody (Fig. 3C). Pentraxin 3 was processed similarly by MMP-1 and to a lesser extent by MMP-2 (see Fig. S3C in the supplemental material). Deficiency of NPC2, a cholesterol-binding protein involved in lysosomal storage of cholesterol and other lipids, causes Niemann-Pick C2 disease, a fatal neurodegenerative disorder (70). Recombinant human NPC2 was processed from an apparent molecular mass of 17.8 kDa by MMP-14 to at least five forms with lower molecular masses

4904

BUTLER ET AL.

ranging from 14.8 to 7.4 kDa (Fig. 3D). Iduronate-2-sulfatase, which participates in glycosaminoglycan metabolism and a deficiency of which manifests as the lysosomal storage disorder Hunter disease (54), was processed from an apparent molecular mass of 97 kDa to fragments of 57.5 kDa and 31.6 kDa by MMP-14 (Fig. 3E). These processed fragments migrate with the more diffuse autodegraded MMP-14 (Fig. 3B, MMP-14 control lane) but can be seen as discrete bands. Iduronate-2sulfatase was also processed, at higher efficiency, by MMP-2 and MMP-8 (see Fig. S3D in the supplemental material). Hence, a number of proteins that were implicated by proteomic analysis as being shed by MMP-14, based on increased levels in the conditioned medium upon expression of MMP-14 in MDA-MB-231 cells and decreased levels in the presence of a MMPI, were biochemically validated as substrates of MMP-14 in vitro. However, this was not the case for all the proteins tested. Although the MMPI/vehicle ICAT ratios of the protease inhibitors elafin, Kunitz-type protease inhibitor 1, and tissue inhibitor of metalloproteinase 1 (TIMP-1) were decreased, the elafin and Kunitz-type protease inhibitor 1 proteins were not significantly cleaved by MMPs in vitro (data not shown), and TIMP-1 is a specific MMP inhibitor, though it does not inhibit MMP-14 (141). Thus, changes in the ICAT ratios for these proteins are likely due to indirect effects, such as MMPI modulation of the protease web (91, 92), or perhaps these proteins are present in the conditioned medium secretome only when bound to proteins which are themselves reduced in amount following decreased shedding upon MMPI treatment (Fig. 1B and D). Accumulation of substrates in cell membranes upon MMPI treatment. As well as detecting changes in the levels of shed ectodomains in the conditioned media, we examined membrane preparations from cells incubated in the presence and absence of inhibitor to determine whether the decrease in ectodomain shedding to the conditioned medium correlated with an increase in the protein levels on the cell membrane (see Table S2 in the supplemental material for a complete list of proteins and peptides identified in two separate experiments). Many proteins had MMPI/vehicle ICAT ratios that decreased in the conditioned medium and increased in the membrane preparations (Table 4 highlights several examples, and every peptide identified and ICAT ratio determined for these proteins is presented in Table S7 in the supplemental material). These included single-pass type I and type II membrane proteins (e.g., Axl receptor tyrosine kinase and catecholO-methyltransferase), multipass membrane proteins (e.g., chloride intracellular channel protein 1, SERCA2), and glycophosphatidylinositol-anchored proteins (e.g., CD59 and uPAR) for which a direct shedding activity can be visualized. Some of the proteins are not themselves membrane proteins but are likely to be bound to the cell via interactions with membrane-tethered molecules such as heparan sulfate proteoglycans and receptors (Fig. 1B and D) or by interaction with exosites on the stabilized inhibited mature MMP-14, a form of “substrate trap” (Fig. 1E). Western blotting was carried out to confirm the ICAT ratios and to validate the MMPI-dependent reduction in shedding of Axl receptor tyrosine kinase and CD59 (Fig. 4). The amount of the 68.2-kDa (apparent molecular mass) ectodomain of Axl receptor tyrosine kinase, a receptor for Gas6 which stimulates

MOL. CELL. BIOL. TABLE 4. Cell membrane accumulation of membrane or membrane-associated substrate candidatesa MMPI/vehicle Protein

Conditioned medium and membranes hnRNP E1 MMP-1b Chloride intracellular channel protein 1 Axl receptor tyrosine kinase RCC2 protein uPARb hnRNP R Multifunctional protein ADE2 CD59

Medium Ratio

Ratio

No. of peptides

0.51 0.71 0.74

2 2 5

1.66 1.53 1.53

2 1 1

0.24 0.57 0.58 0.69 0.71

1 1 1 3 2

1.45 1.39 1.36 1.29 1.28

1 1 1 3 1

0.41

2

1.14

2

1.73 1.71 1.62 1.61 1.29 1.27

2 2 1 1 1 1

1.17

4

Membranes only Catechol O-methyltransferase SERCA2 Coatomer subunit VDAC-2 Myoferlin Na⫹/K⫹-transporting ATPase ␣-4 chain Integrin ␣3b Conditioned medium only TNF receptor superfamily member 12A CD166 (ALCAM Melanotransferrin

Membrane

No. of peptides

0.31

1

0.61 0.75

1 1

a Proteins with increased ICAT ratios in the membrane fraction (Membrane) of MMP-14-transfected MDA-MB-231 cells and known membrane proteins with ratios decreased in the conditioned medium (Medium) in the presence of the inhibitor are implicated as being shed by metalloproteinases (MMPI/vehicle). Also shown are (i) membranes only, i.e., membrane proteins increased in the presence of MMPI, although peptides were not detected in the conditioned medium, and (ii) conditioned medium only, i.e., peptides for ectodomains of membrane proteins that showed decreases in conditioned medium but were not detected in the membrane preparation. Peptides were identified at ⱖ99% confidence, unless otherwise noted in Tables S1 and S2 in the supplemental material. Values shown are averages of all the peptide ratios quantified for each experiment. The individual peptide sequences are shown in Table S7 in the supplemental material. Abbreviations: hnRNP, heterogeneous nuclear ribonucleoprotein; SERCA2, sarcoplasmic/endoplasmic reticulum calcium ATPase 2; VDAC-2, voltage-dependent anion-selective channel protein 2; TNF, tumor necrosis factor; ALCAM, activated leukocyte cell adhesion molecule. b Known substrate.

cell proliferation (78), in the conditioned medium was reduced in the presence of the MMPI, consistent with the ICAT MMPI/ vehicle ratio of 0.24 (Fig. 4A). CD59, a glycophosphatidylinositol-anchored inhibitor of complement membrane attack complex formation (126), was decreased in the medium and increased in the membrane fraction in the presence of the MMPI, consistent with the MMPI/vehicle ICAT ratios of 0.41 for the medium and 1.14 for the membrane sample (Fig. 4B). Mechanistically, we examined this further. Shedding of these two membrane proteins was enhanced by the treatment of MDA-MB-231 cells with 12-O-tetradecanoylphorbol-13-acetate, which is known to upregulate cell surface expression of various proteases, including MMP-14 (71) (Fig. 4C and D), consistent with the MMP-14-transfected cell results.

VOL. 28, 2008

PHARMACOPROTEOMICS OF A METALLOPROTEINASE INHIBITOR

4905

FIG. 4. Regulation of ectodomain shedding of Axl receptor tyrosine kinase and CD59 by MMPI and 12-O-tetradecanoylphorbol-13-acetate (TPA). Conditioned medium or membrane fractions (5 ␮g total protein) from MMP-14-transfected MDA-MB-231 cells treated with 10 ␮M prinomastat (⫹MMPI) or vehicle (⫺MMPI) were analyzed by SDS-PAGE and Western blotting. (A) Axl receptor tyrosine kinase (Axl) in conditioned medium was detected on 10% polyacrylamide gels, using an antibody specific for the Axl receptor tyrosine kinase ectodomain. (B) CD59 was detected on 13% polyacrylamide gels, using the monoclonal antibody BRIC 229. Panels show results from a single Western blot, but the lower panel is overexposed to show CD59 in the medium sample. Blots were developed with appropriate horseradish peroxidase-conjugated secondary antibodies and enhanced chemiluminescence. Positions of molecular mass markers as 103 Da are shown. Conditioned medium or cell lysates (5 ␮g total protein) from MDA-MB-231 cells (transfected with empty vector) treated with 0.2 ng/ml TPA (⫹) or untreated (⫺) under serum-free conditions for 72 h were subjected to reducing SDS-PAGE and Western blotting. Axl receptor tyrosine kinase (C) and CD59 (D) were detected as described above.

The MMPI/vehicle ICAT ratio for uPAR was decreased in the conditioned medium (0.58) and increased in the membrane fraction (1.36) of MDA-MB-231 cells expressing MMP-14 (Table 4). This indicates that shedding had been inhibited by the MMPI. uPAR is cleaved by MMP-3, -12, -19, and -25 in the linker between domains D1 and D2 to release the D1 domain (4, 60). However, the peptide detected for uPAR in the conditioned medium, 269SGCNHPDLDVQYR281, is at the C terminus of the D3 domain, proximal to the plasma membrane (see Fig. S4A in the supplemental material), suggesting that soluble uPAR (suPAR), composed of domains D1, D2, and D3, is shed. We tested peptides encompassing the D3 domain juxtamembrane C-terminal peptide (272NHPDLDVQYRSG283) or the known MMP cleavage site between D1 and D2 (81SGRAVTYSRSRYLEC95) as a control for cleavage to determine whether MMPs might shed suPAR. MMP-2 and MMP-8, but not soluble MMP-14, processed the D1-D2 peptides at a site equivalent to the T86-Y87 site in uPAR, as reported for MMP-3, -12, -19, and -25 (7), as well as at the Y92-L93 site (see Fig. S4B in the supplemental material). However, it is unclear whether this site in the intact protein might be cleavable, as the Cys in P3⬘ is likely be disulfide cross-linked, a state that might prevent cleavage at site Y92-L93. Chymotrypsin and plasmin also cleaved this peptide at the reported plasmin site (R89-S90) (7). However, neither MMP-14 nor any of the other MMPs tested cleaved the D3 peptide, which was cleaved by both

plasmin and chymotrypsin, as reported previously (7). Although MMP exosite interactions with sites present in the uPAR protein or interactions with other binding proteins might be required for cleavage, the peptide cleavage data suggest that neither MMP-14 nor MMP-1, -2, -8, or -9 directly shed suPAR. Identification of novel MMP-14 substrates. Identification of such a large number of known MMP substrates in the conditioned medium (Table 2) reveals the dynamic nature of cell surface proteolysis and indicates that the administration of an MMPI enables a successful degradomic screen that can reveal substrates, both known and novel. In particular, the blockade of metalloproteinase cleavage, shown by the reversal of the ICAT ratios obtained for MMP-14/vector (no MMPI) compared with those of MMPI/vehicle (Table 3), demonstrates that the combined analysis of cells, ICAT labeling, and MS-MS in the absence of inhibitor and then following inhibitor treatment can be used reliably to identify proteolysis-related changes in the proteome. Moreover, upon administration of a protease inhibitor drug, the reversal of ICAT ratios allows accurate prediction of substrates with high confidence and reveals drug-induced changes valuable for drug development. Therefore, a number of proteins which exhibited MMPI/vehicle ratios of ⱕ0.77 in the conditioned medium, i.e., within the range observed for known MMP substrates, were also selected for validation that might be novel substrates of MMP-14

4906

BUTLER ET AL.

MOL. CELL. BIOL.

TABLE 5. Candidate metalloprotease substrates identified in conditioned medium of MMP-14-transfected MDA-MB-231 cells that show reduced shedding in the presence of the MMPIa Candidate substrate

Progranulin EGF-containing fibulin-like extracellular matrix protein 1 IGFBP-7 Proprotein convertase subtilisin/ kexin type 9 Cyr61 Serine protease 23 Niemann-Pick disease, type C2 variant KIAA1392/Storkhead-box 2 IGFBP-4 Kunitz-type proteinase inhibitor 2 KIAA0230/peroxidasin homolog Beta-2-microglobulin Dickkopf-1 Extracellular matrix protein 1 Pentraxin-related protein PTX3 Galectin-3-binding protein Elafin Thrombospondin-1 Cathepsin A Cathepsin B Legumain Lupus La protein Peptidyl-prolyl cis-trans isomerase B Macrophage migration inhibitory factor Peptidyl-prolyl cis-trans isomerase A Gamma enolase DJ-1 Macrophage capping protein Quiescin Q6 Stanniocalcin-1

MMPI/vehicle ratio

No. of peptides

0.07 0.25

1 1

0.26 0.28

3 1

0.29 0.36 0.36

10 1 4b

0.39 0.40 0.44 0.45 0.47 0.48 0.50 0.51 0.51 0.53 0.54 0.55 0.57 0.59 0.59 0.64 0.67

1 6 1 5 1 3 6b 1 1 2 2 1 3 3 1 1 1

0.68 0.68 0.69 0.70 0.72 0.75

4b 3 3 1 5 3

and C-terminal DJ-1 fragments of 13.5, 8.7, 7.9, 7.0, and 5.4 kDa (Fig. 5B). The N-terminal sequences shown were determined by Edman degradation. The large digestion products (40.7 to 23.4 kDa) were N-terminal fragments commencing with MSPIL, the N terminus of GST. Since GST is 26.4 kDa, MMP-14 cleavage occurred at more than two sites within DJ-1. Cleavage by MMP-14 also gave fragments with N-terminal sequences of LAGKD and LLAHE, corresponding to cleavage in DJ-1 at positions 37G-38L and 111A-112L, respectively (Fig. 5C). Other cleav-

a MDA-MB-231 breast cancer cells transfected with MMP-14 were treated with the MMPI prinomastat or with DMSO vehicle under serum-free conditions for 48 h. Proteins in each conditioned medium were labeled with ICAT reagents (heavy or light), pooled, and digested with trypsin. The biotin-tagged tryptic peptides were separated by multidimensional LC and analyzed by MS-MS. Values shown are averaged abundance ratios from three separate experiments (MMPI/vehicle) for peptides detailed in Table S8 in the supplemental material. Peptides were identified at a ⱖ99% confidence level, unless otherwise noted in Tables S1 and S8 in the supplemental material. Abbreviations: IGFBP, insulinlike growth factor binding protein. b Peptide numbers include those differing only by the oxidation of a methionine, which were counted as two peptides, since they were identified independently in the MS analysis.

(Table 5 shows selected proteins; Table S8 in the supplemental material for a complete list of individual peptides and ICAT ratios for these proteins). Several of these proteins were assayed for cleavage in vitro to biochemically validate their identification as novel substrates and to characterize their cleavage by MMP-14. DJ-1 has been ascribed several different functions, and mutations in DJ-1 predispose to early-onset Parkinson’s disease (66). The recombinant GST-DJ-1 fusion protein was processed efficiently by MMP-14 in a concentration-dependent manner (Fig. 5A). At a 1:500 enzyme/substrate molar ratio, most of the GSTDJ-1 fusion protein was converted from an apparent molecular mass of 54.9 kDa to N-terminal fragments of 40.7, 33.9, 25.7, and 23.4 kDa, composed of the GST and N-terminal portions of DJ-1,

FIG. 5. DJ-1 is a substrate of MMP-14. (A) GST-DJ-1 (435 pmol) was incubated for 18 h at 37°C with increasing concentrations of MMP-14 (GST-DJ-1/MMP-14 molar ratios of 1:0, 1,000:1, 500:1, 250:1, 100:1, and 50:1), or MMP-14 was incubated alone (in amounts equivalent to a 100:1 ratio). (B) GST-DJ-1, digested with 500:1 MMP14, was blotted onto polyvinylidene difluoride membrane. Arrows indicate bands subjected to Edman sequencing. Apparent molecular weights and sequences obtained for the first five residues are shown. Samples were electrophoresed on 12.5% (A) and 15% Tris-Tricine (B) SDS-polyacrylamide gels and stained with Coomassie brilliant blue R250. Molecular mass markers as 103 Da are shown. (C) Schematic diagram of GST-DJ-1. The start and end of the pGEX-5X-1 vector fusion protein sequence are shown, followed by the entire DJ-1 sequence. N-terminal sequences identified by Edman degradation are underlined. Cleavage sites within DJ-1 are indicated by arrowheads and those in the GST fusion protein by open arrowheads.

VOL. 28, 2008

PHARMACOPROTEOMICS OF A METALLOPROTEINASE INHIBITOR

age sites are discussed in Results in the supplemental material. Hence, DJ-1 is efficiently processed at multiple sites by MMP-14 and by MMP-1, -2, -8, and -9 (see Fig. S3E in the supplemental material), and this may play a role in the pathogenesis of Parkinson’s disease. We validated the MMPI/vehicle ICAT ratio of 0.54 for TSP-1, using Western blotting (Fig. 6A). TSP-1 levels were decreased in the conditioned medium by the MMPI compared with those in the vehicle, confirming the MMP-dependent shedding. TSP-1 was also processed by MMP-14 (Fig. 6B), as well as by other MMPs (data not shown). Peptidyl-prolyl cis-trans isomerase A (PPI-A, also termed cyclophilin A) is an intracellular molecular chaperone that, when secreted, functions as a proinflammatory cytokine, signaling via CD147 (EMMPRIN) and activating endothelial cells (55, 144). PPI-A (18 kDa) was proteolyzed by MMP-14 to two fragments of 13.9 and 4.2 kDa that were detected by Western blotting with an anti-PPI-A polyclonal antibody and were associated with a reduced amount of the full-length immunoreactive protein (Fig. 6C). The dickkopf-1 protein, a wnt antagonist (28), undergoes proteolysis near its N terminus by MMP-14 and shows an apparent molecular mass shift from 35.5 kDa to a 31.6-kDa fragment, which retains the immunoreactive C-terminal His tag (Fig. 6D). Gamma enolase (neuron-specific enolase), the ␥␥ form of the glycolytic enzyme enolase (57), was proteolyzed by MMP14, but cleavage products (ranging from apparent molecular masses of 49.0 to 33.9 kDa) were faint and disappeared at higher MMP-14 concentrations, suggestive of complete degradation (Fig. 6E). However, discrete fragments were produced by MMP-2, -9, -1, and -8 (see Fig. S3F in the supplemental material), highlighting the efficiency of MMP-14 proteolysis. The connective tissue growth factor (CTGF), cysteine rich protein (Cyr61), and nephroblastoma overexpressed (NOV) gene (CCN) family (99) includes CTGF, a known MMP substrate (25, 43). Cyr61 promotes proliferation, chemotaxis, angiogenesis, and cell adhesion. Cyr61 was converted from an apparent molecular mass of 45 kDa to a major fragment of 21.9 kDa and a minor one of 11.0 kDa (Fig. 6F). Progranulin is a multifunctional glycoprotein involved in development, tumorigenesis, inflammation, and repair that contains granulins 1 to 7, which are released by proteolysis by elastase (45). Progranulin was processed by MMP-14 from an apparent molecular mass of 67.6 kDa into several lower-molecular-mass forms of 50.1, 39.8 (T18RCPD, N terminus), 30.2 (L360KRDV), 26.9 (L360KRDV), and 21.9 (T18RCPD) kDa, as observed with Western blotting with an anti-progranulin polyclonal antibody (Fig. 6G, N-terminal sequences determined by Edman degradation are bracketed). MMP-14 cleaved progranulin at site A359-L360, between granulin 4 and granulin 5. Predicted masses for residues 18 to 359 and 360 to 593 are 36.3 and 25.4 kDa, respectively, thus, further cleavages may occur. Beta-2-microglobulin and macrophage migration inhibitory factor were not processed by MMP-14 or MMP-1, -2, -8, and -9 in vitro (data not shown), indicating that levels of these proteins are altered due to downstream effects. DISCUSSION We have demonstrated that a pharmacoproteomic screen used to examine the effects of a protease inhibitor drug on cells can successfully predict protease substrates, as well as identify key molecules and pathways that are altered by protease expression

4907

and upon drug administration. This study profiles the widespread effects of metalloproteolytic activity on the secretome of human breast carcinoma cells and reveals the effects of blocking this activity by using a protease inhibitor drug. This approach has also unveiled a multitude of direct and indirect effects on the proteome that may be useful for predicting side effects of drug treatment in patients. In so doing, a rich diversity of proteins were found to be proteolytically shed from the plasma membrane. By using MMP-14 transfection, 12-O-tetradecanoylphorbol-13-acetate stimulation, and MMPI blockade, we showed the dynamic nature of proteolytic shedding and that the accumulation of cleaved ectodomains and protein fragments was metalloproteinase dependent, executed by the cell surface membrane type 1 MMP, MMP-14. In nonbiased quantitative proteomic approaches, criteria must be set to identify which of the myriad proteins whose levels are modulated by a protease are substrates and which are indirect effects wrought by an altered cellular signaling environment due to proteolytic processing of individual components of signaling pathways or ripples in the protease web. We chose to set a “cutoff” value based upon ratios for known substrates in this study. These values were less extreme than one might set using arbitrary cutoff values, but they reflect the levels of change that occur in a complex system. Some novel substrates can be inferred, since many proteins are substrates of more than one MMP due to homology and redundancy of this protease family. Likewise, molecules from families in which other members are processed by MMPs are strong candidate substrates. The IGFBP family members IGFBP-1, -3, -5, and -6 are cleaved by several MMPs (Table 2) to modulate the bioavailability of insulin-like growth factors. This and the decreased MMPI/vehicle ICAT ratios are good indicators that IGFBP-7 and IGFBP-4 are also MMP substrates. CTGF (CCN2) is a known substrate of MMPs and the related Cyr61 (CCN1) was biochemically confirmed as a MMP-14 substrate in this study. MMPs may be key regulators of the galectin family, carbohydrate binding proteins that regulate cell survival (49). We have demonstrated that galectin-1, a potential drug target due to its involvement in tumorigenesis (103) and inflammation (109), is a novel substrate of MMP-14 in vitro. Galectin-1 was recently shown to be a substrate for MMP-2 (26), and galectin-3 is cleaved by MMP-14, -2, and -9 (84, 85, 134). Here, candidate substrates were identified from a cellular environment, which supports the concept that they are indeed natural MMP substrates. For some novel MMP-14 substrates, such as DJ-1, processing to fragments in vitro was complete; for others it appeared less efficient. A relatively poor turnover of a substrate in vitro may reflect a deficiency of cofactors, binding partners, and interactions that were present in the cellular milieu and which increase the efficiency of processing in vivo. Conversely, just because an enzyme cleaves a protein in vitro does not mean that it will do so in vivo (10). For example, fibronectin is efficiently cleaved by many MMPs, including MMP-14, in vitro, but previously (129), we showed in a cell system that it is shed but not proteolyzed by MMP-14. In contrast, MMP-2 in a similar cell-based system degraded fibronectin (26), as reflected by isotope-labeled peptide ratios that were the opposite (⬍1) of those in MMP-14-expressing systems (⬎1) (129). This likely reflects the different partition-

4908

BUTLER ET AL.

MOL. CELL. BIOL.

VOL. 28, 2008

PHARMACOPROTEOMICS OF A METALLOPROTEINASE INHIBITOR

ing of these two proteases with respect to the substrate, to the cell membrane (MMP-14) and the secretome (MMP-2), emphasizing the need for cell-based analyses of proteolysis to determine biological relevance. In vitro lipopolysaccharide-induced CXC chemokine (LIX) is cleaved at position 4-5 by MMP-1, -2, -8, -9, -12, -13, and -14, increasing bioactivity via its cognate receptor CXCR2 (131). However, neutrophil infiltration toward lipopolysaccharide is almost entirely abrogated in Mmp8⫺/⫺ mice, demonstrating a lack of physiological redundancy in vivo (131). Thus, after a candidate substrate is identified by proteomic screening, validation is required to confirm processing in vivo, to determine the enzyme(s) responsible, and to characterize the functional consequences of proteolytic processing. While a change in the levels of a protein in the presence of MMPI compared with those of a vehicle is an indication that the protein may be a substrate, levels may also change due to indirect effects. These effects include release of a protein interactor of the processed protein or proteoglycan; the effects on a cascade in the protease web, for example, activation of a second protease by MMP-14, such as MMP-13 (59) or MMP-2 (112, 125), which then cleaves the substrate; altered signaling and hence transcriptional events; or inhibition of other metalloproteases, such as members of the ADAM/ADAM-TS families due to the broad specificity profile of some MMP-directed hydroxamate inhibitors. Beta-2-microglobulin, elafin, Kunitztype protease inhibitor 1, cystatin C, GRO␣, follistatin-related protein 1, and uPAR exhibited altered MMPI/vehicle ICAT ratios but did not appear to be processed by MMPs in vitro. Elafin binds to extracellular matrix proteins via transglutaminase cross-linking mediated by its N-terminal domain (114). Thus, shedding of this inhibitor bound to the actual MMP substrate is likely as elafin is also resistant to MMP-8 (48). This has been described for the chemokine KC, which binds to syndecan-1 (67), and peptidyl-prolyl cis-trans isomerase B (cyclophilin B), which binds heparan sulfate proteoglycans (2, 27) and which was also decreased in the conditioned medium from the MMPI-treated cells (MMPI/vehicle ICAT ratio, 0.64 [Table 5]). The protease responsible for shedding suPAR (uPAR extracellular domains D1 to D3), which is increased in tumor and phorbol ester-treated cells, is the subject of much interest.

4909

The peptide detected in conditioned medium for uPAR suggested shedding of suPAR, but we found that MMPs were unable to cleave a peptide containing the suPAR cleavage site. However, suPAR can be released by plasmin and uPA by cleavage at site R303 to S304 at the cell membrane (7). The MMPI/vehicle ICAT ratio for uPA was 0.43, indicating that there was decreased uPA in the conditioned medium. This in itself, as well as less conversion of plasminogen to plasmin, could reduce cleavage of uPAR. Alternatively, a peptide mimic of the cleavage site might be insufficient for cleavage by MMPs if interaction with exosites or other binding proteins is required. Although these indirect effects may not be useful in terms of a degradomic screen, they are important biologically and critical in terms of drug validation and therefore require characterization. Once again, this highlights the need for a system-wide approach to understand proteolysis and drug treatment in the broadest context. MMPs are now recognized as processors of a wide range of signaling molecules and bioactive mediators (17, 91). This is exemplified by the variety of known MMP substrates identified, which included chemokines (GRO␣, IL-8), growth factor binding proteins (IGFBPs, CTGF, TGF-beta binding protein-1S), cell surface receptors (uPAR, gC1qR, integrins), enzymes (MMP-1, uPA), and proteinase inhibitors (tissue factor pathway inhibitor, cystatin C), as well as extracellular matrix (ECM) proteins (e.g., collagen [VI], laminin, fibrillin). The decrease of these proteins in the conditioned medium of MMP-14-transfected MDA-MB-231 cells with prinomastat versus vehicle implicates MMP-14 in their processing and release. Similarly, other proteins and novel substrates that were modulated by the MMPI are diverse, encompassing ECM proteins (epidermal growth factor-containing fibulin-like ECM protein 1, TSP-1, ECM-1), innate immunity and inflammatory mediators (pentraxin 3, peptidyl-prolyl cis-trans isomerase A), receptors (Axl receptor tyrosine kinase, CRIM-1), proteases (cathepsins A and B, proprotein convertase subtilisin/ kexin type 9, serine protease 23, legumain), and protease inhibitors (elafin, Kunitz-type proteinase inhibitor 2). Since the proteins are so varied that are affected by the prinomastat treatment, their modulation will likely have diverse effects. Hence it is not surprising that MMPI clinical trials to inhibit MMP-directed cleavage of basement mem-

FIG. 6. Validation of candidate MMP substrates identified in conditioned medium. (A) Conditioned medium (10 ␮g total protein) from MMP-14-transfected MDA-MB-231 cells treated with vehicle (⫺MMPI) or 10 ␮M prinomastat (⫹MMPI) (nonreduced) was separated by 12% SDS-PAGE, and TSP-1 (TSP-1) was detected by Western blotting and enhanced chemiluminescence using the mouse monoclonal antibody MAII, which recognizes the heparin binding domain of TSP-1. (B) Recombinant TSP-1 was incubated with or without MMP-14, and cleavage products were analyzed by SDS-PAGE on 9% gels by silver staining. Fragments are indicated by arrows. (C) Samples of peptidyl-prolyl cis-trans isomerase A (PPI-A) incubated with or without MMP-14 were analyzed on 15% Tris-Tricine SDS-polyacrylamide gels, Western blotted with a rabbit anti-PPI-A polyclonal antibody. (D) Recombinant dickkopf-1 with a C-terminal His tag incubated with increasing concentrations of MMP-14 (1:50, 1:10, and 1:5 molar ratio enzyme/substrate) was analyzed on 12.5% Tris-Tricine SDS-polyacrylamide gels. Western blotting was carried out with an anti-polyhistidine antibody. (E) Gamma enolase was incubated with increasing concentrations of MMP-14 (1:1,000, 1:500, 1:250, 1:100, 1:50, and 1:10 enzyme/substrate molar ratio), or MMP-14 was incubated alone (equivalent to a 1:10 ratio). Products were analyzed on 12.5% Tris-Tricine SDS-polyacrylamide gels by silver staining. (F) Cyr61 cleavage fragments produced upon incubation with MMP-14 were visualized on 15% Tris-Tricine SDS-polyacrylamide gels by silver staining. (G) Progranulin processing by MMP-14 was visualized on 12.5% Tris-Tricine SDSpolyacrylamide gels, Western blotted with a rabbit polyclonal antibody raised against progranulin. Fragments are shown by arrows, with N-terminal sequences obtained by Edman degradation. The sequence of progranulin is shown in the bottom panel: residues 1 to 17 constitute the signal sequence which is removed; residue T18 is the mature N terminus of progranulin; constituent granulins are boxed; N-terminal sequences of fragments are underlined; peptides identified by ICAT are in bold; the MMP-14 cleavage site is indicated by an arrow. All recombinant proteins were incubated for 18 h at 37°C with MMP-14. Western blots were detected using species-appropriate Alexa-Fluor 680-conjugated secondary antibodies (Molecular Probes) on an Odyssey infrared scanner (LiCor), unless otherwise stated. Arrows indicate cleaved protein fragments, and apparent molecular weights and masses are shown. Positions of molecular mass markers as 103 Da are indicated.

4910

BUTLER ET AL.

brane proteins to prevent metastasis were unsuccessful: even with a current understanding of the expansive MMP signaling degradome, it would be impossible to predict the wide range of proteins and pathways that are affected by the MMPI drug. Side effects of the MMPIs in clinical trials, e.g., tendonitis and myalgia, suggested inhibition of homeostatic tissue turnover, leading to fibrosis and inflammation (30, 132). Several of the substrates and candidate substrates identified in this pharmacoproteomic screen might be responsible; for example, dickkopf-1 has been shown to orchestrate joint remodeling (28), and the CCN family members CTGF and Cyr61 are implicated in pathologies characterized by ECM deposition and fibrosis, myofibroblast formation, and chemotaxis (19, 20). Notably, myofibroblasts were found in higher amounts in tendon sites in MMPI-treated rats and marmosets. The potential role of MMP processing in these functions is currently under study. Proteases are not only responsible for protein degradation but, by limited and specific cleavages, can act as switches, turning protein activity on or off, or they can modulate protein function in more complex ways. Processing can turn an agonist into an antagonist, inactivate an inhibitor, increase the bioavailability of a growth factor, convert receptor ectodomains into soluble binding proteins, and reveal fragments with new functions. For example, MMP cleavage of just four amino terminal residues of chemokines that are expressed in inflammation converts agonists to antagonists (77), activates other chemokines (131), converts CXCL12 to a neurotoxin that switches receptor specificity (145), and sheds cell membranebound fractalkine (26). MMPs are critical regulators of cellular functions that orchestrate every stage of tumorigenesis, including apoptosis, growth, angiogenesis, metastasis, and innate immunity (18, 92). Proteolytic cleavage can abrogate, exacerbate, or create new functions that may aid or impede a cancer therapeutic regimen. TSP-1 is a case in point: TSP-1 is a large modular molecule with multiple domains, ligands, and receptor binding sites (118). The effects of TSP-1 on angiogenesis and tumorigenesis are somewhat controversial. TSP-1 has been used for anticancer therapeutic trials, using both the whole molecule (3, 106) and modules such as the TSP-1 repeats (146, 147) or the peptides thereof (110), which are antiangiogenic. Other modules such as the CD47 binding domain counteracted chemotherapy by inhibiting apoptosis (104). As implicated in our study, proteolytic processing of TSP-1 could regulate the modules which are available for interaction and hence modulate the overall effect of TSP-1 on tumorigenesis. Other novel substrates have domains which could be proteolytically processed to modify activity; e.g., as well as being intracellular, DJ-1 is present in biological fluids such as serum and cerebrospinal fluid, secreted by cultured melanoma cells, and is overexpressed by some human tumors (65, 82, 96, 139). Reported activities include the regulation of transcription, PTEN tumor suppression, apoptosis, oncogenesis, molecular chaperone activity, and protection against oxidative stress (12). Mutations in DJ-1 are implicated in recessive, early-onset Parkinson’s disease, perhaps due to a loss of neuroprotective antioxidant activity. DJ-1 contains a putative active site similar to that of cysteine proteases, though the catalytic triad is orientated unfavorably, with C-terminal helix 9 blocking the putative catalytic site (53), and only a weak activity against a fluorogenic casein substrate has been reported (86). Proteolytic

MOL. CELL. BIOL.

processing of DJ-1 by MMPs might remove this C-terminal regulatory region to activate proteolytic activity. However, no activity was detected against resorufin-labeled casein, a general protease substrate, following processing of GST-DJ-1 by MMP-14 (data not shown). Neuron-specific (gamma) enolase is expressed in breast carcinomas (47) and is a serum and cerebrospinal fluid marker for neurological damage (115). The neurotrophic and neuroprotective activities of neuron-specific enolase lie in the C-terminal 30 residues (44), and these functions could be modified by the MMP processing reported here. Progranulin regulates development, repair, and cancer progression (45). This molecule contains within its sequence granulins 1 to 7, 6-kDa peptides released by elastase proteolysis, which have independent functions, for example promoting proliferation and inflammation (45). The existence of intermediate cleaved forms which may have unique functions is suggested as 25-kDa epithelial transforming growth factor was found to have the same N terminus as that of granulin 4 (98), and here, cleavage between granulins 4 and 5, at site A359-L360 generated 30- to 40-kDa products. A large number of candidate substrates have reported roles in or significance to cancer; for instance, of the 30 candidate substrates shown in Table 5, 24 have cancer-related references in the literature (see Table S9 in the supplemental material); thus, proteolytic regulation of these and modulation by inhibitor drugs may significantly impact cancer therapy. Targeting a protease might have unexpected positive or negative influences on a disease course, depending upon whether the enzyme is a drug target or an antitarget (92). For instance, contrary to the findings of all previous studies with MMPdeficient mice, Mmp8⫺/⫺ mice demonstrated enhanced neutrophil accumulations, rather than neutrophil infiltration that was hampered due to decreased collagenolysis as hypothesized (131), as well as augmented susceptibility to chemically induced skin tumors and arthritis (J. H. Cox and C. M. Overall, unpublished data). Thus, MMP-8 became the first MMP antitarget in cancer therapy research (6). Some proteins that have altered shedding in the presence of the MMPI are also likely to be cancer antitargets (92). Blockade of shedding of these proteins might be detrimental to an anticancer strategy; for example, CD59 protects cells by inhibiting the formation of the complement membrane attack complex, and this is exploited by both viruses and tumor cells which overexpress CD59 to escape complement-mediated killing (56). CD59 is also implicated in calreticulin binding and signaling (58), as well as in promoting tumor angiogenesis (136), as are IGFBP-7, vimentin, and high-mobility group box 1, levels of which were also modulated by MMP-14 expression or MMPI treatment (see Tables S1, S2, and S5 in the supplemental material). Inhibiting shedding of other molecules may be beneficial; e.g., beta-2microglobulin is a component of the major histocompatibility complex class I (MHC-I) complex, which may be shed by tumor cells (though probably not by direct processing, since it was not a MMP substrate in vitro) to prevent recognition of tumor antigens by CD8⫹ T cells, thus escaping immune detection (1, 105). MMPs also have roles which are host protective, for instance, the generation of the angiogenesis-blocking neopeptides angiostatin and vasostatin from plasminogen and calreticulin, respectively (87, 100). It is clear that the substrate degradome of MMPs and other

VOL. 28, 2008

PHARMACOPROTEOMICS OF A METALLOPROTEINASE INHIBITOR

proteases must be defined during the validation of proteases as drug targets, to aid in the decision to treat disease using antiprotease drugs. Certainly, this could help to predict and minimize the side effects caused by long-term antiprotease drug administration to patients and so also aid in medicinal chemistry modification of drug leads during drug development. Since MMP inhibitors are not yet specific for a single MMP and since the blanket inhibition of MMPs is almost certain to cause side effects due to the many substrates, each with particular biological functions, it makes sense to target particular substrates of MMPs whose function is critical for tumorigenesis or metastasis. This requires a concerted effort at MMP substrate discovery using degradomic studies which will be invaluable for the development of specific and effective anticancer drugs. Identification of protease substrates should also provide leads for selecting new biomarkers of disease, since many biomarkers are stable proteolytic fragments (68). Thus, rather than selecting biomarkers based on changes in concentration determined by using enzyme-linked immunosorbent assay, for example, new biomarkers might be discovered by determining the proteolytically processed state of the proteome in normal versus disease states.

12.

13.

14.

15.

16.

17.

18. 19. 20.

ACKNOWLEDGMENTS C.M.O. is supported by a Canada Research Chair in Metalloproteinase Proteomics and Systems Biology. This work was supported by research grants from the Canadian Institutes of Health Research, the National Cancer Institute of Canada (with funds raised by the Canadian Cancer Society), and the Canadian Breast Cancer Research Alliance Special Program Grant on Metastasis, as well as with an infrastructure grant from the Michael Smith Research Foundation. REFERENCES 1. Abdul, M., and N. Hoosein. 2000. Changes in beta-2 microglobulin expression in prostate cancer. Urol. Oncol. 5:168–172. 2. Allain, F., C. Vanpouille, M. Carpentier, M. C. Slomianny, S. Durieux, and G. Spik. 2002. Interaction with glycosaminoglycans is required for cyclophilin B to trigger integrin-mediated adhesion of peripheral blood T lymphocytes to extracellular matrix. Proc. Natl. Acad. Sci. USA 99:2714–2719. 3. Allegrini, G., F. A. Goulette, J. W. Darnowski, and P. Calabresi. 2004. Thrombospondin-1 plus irinotecan: a novel antiangiogenic-chemotherapeutic combination that inhibits the growth of advanced human colon tumor xenografts in mice. Cancer Chemother. Pharmacol. 53:261–266. 4. Andolfo, A., W. R. English, M. Resnati, G. Murphy, F. Blasi, and N. Sidenius. 2002. Metalloproteases cleave the urokinase-type plasminogen activator receptor in the D1–D2 linker region and expose epitopes not present in the intact soluble receptor. Thromb. Haemostasis 88:298–306. 5. Ashworth, J. L., G. Murphy, M. J. Rock, M. J. Sherratt, S. D. Shapiro, C. A. Shuttleworth, and C. M. Kielty. 1999. Fibrillin degradation by matrix metalloproteinases: implications for connective tissue remodelling. Biochem. J. 340:171–181. 6. Balbin, M., A. Fueyo, A. M. Tester, A. M. Pendas, A. S. Pitiot, A. Astudillo, C. M. Overall, S. D. Shapiro, and C. Lopez-Otin. 2003. Loss of collagenase-2 confers increased skin tumor susceptibility to male mice. Nat. Genet. 35:252–257. 7. Beaufort, N., D. Leduc, J. C. Rousselle, A. Namane, M. Chignard, and D. Pidard. 2004. Plasmin cleaves the juxtamembrane domain and releases truncated species of the urokinase receptor (CD87) from human bronchial epithelial cells. FEBS Lett. 574:89–94. 8. Belaaouaj, A. A., A. Li, T. C. Wun, H. G. Welgus, and S. D. Shapiro. 2000. Matrix metalloproteinases cleave tissue factor pathway inhibitor. Effects on coagulation. J. Biol. Chem. 275:27123–27128. 9. Bini, A., Y. Itoh, B. J. Kudryk, and H. Nagase. 1996. Degradation of cross-linked fibrin by matrix metalloproteinase 3 (stromelysin 1): hydrolysis of the gamma Gly 404-Ala 405 peptide bond. Biochemistry 35:13056–13063. 10. Bini, A., D. Wu, J. Schnuer, and B. J. Kudryk. 1999. Characterization of stromelysin 1 (MMP-3), matrilysin (MMP-7), and membrane type 1 matrix metalloproteinase (MT1-MMP) derived fibrin(ogen) fragments D-dimer and D-like monomer: NH2-terminal sequences of late-stage digest fragments. Biochemistry 38:13928–13936. 11. Bissett, D., K. J. O’Byrne, J. von Pawel, U. Gatzemeier, A. Price, M.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31. 32.

4911

Nicolson, R. Mercier, E. Mazabel, C. Penning, M. H. Zhang, M. A. Collier, and F. A. Shepherd. 2005. Phase III study of matrix metalloproteinase inhibitor prinomastat in non-small-cell lung cancer. J. Clin. Oncol. 23:842– 849. Bonifati, V., B. A. Oostra, and P. Heutink. 2004. Linking DJ-1 to neurodegeneration offers novel insights for understanding the pathogenesis of Parkinson’s disease. J. Mol. Med. 82:163–174. Butler, G. S., R. A. Dean, D. Smith, and C. M. Overall. Membrane protease degradomics: proteomic identification and quantification of cell surface protease substrates. Methods Mol. Med., in press. Butler, G. S., and C. M. Overall. 2007. Proteomic validation of protease drug targets: pharmacoproteomics of matrix metalloproteinase inhibitor drugs using isotope-coded affinity tag labelling and tandem mass spectrometry. Curr. Pharm. Des. 13:263–270. Butler, G. S., D. Sim, E. Tam, D. Devine, and C. M. Overall. 2002. Mannose-binding lectin (MBL) mutants are susceptible to matrix metalloproteinase proteolysis: potential role in human MBL deficiency. J. Biol. Chem. 277:17511–17519. Butler, G. S., E. M. Tam, and C. M. Overall. 2004. The canonical methionine 392 of matrix metalloproteinase 2 (gelatinase A) is not required for catalytic efficiency or structural integrity: probing the role of the methionine-turn in the metzincin metalloprotease superfamily. J. Biol. Chem. 279:15615–15620. Cauwe, B., P. E. Steen, and G. Opdenakker. 2007. The biochemical, biological, and pathological kaleidoscope of cell surface substrates processed by matrix metalloproteinases. Crit. Rev. Biochem. Mol. Biol. 42:113–185. Chabottaux, V., and A. Noel. 2007. Breast cancer progression: insights into multifaceted matrix metalloproteinases. Clin. Exp. Metastasis 24:647–656. Chaqour, B., and M. Goppelt-Struebe. 2006. Mechanical regulation of the Cyr61/CCN1 and CTGF/CCN2 proteins. FEBS J. 273:3639–3649. Chen, C. C., F. E. Mo, and L. F. Lau. 2001. The angiogenic factor Cyr61 activates a genetic program for wound healing in human skin fibroblasts. J. Biol. Chem. 276:47329–47337. Coppock, H. A., A. White, J. D. Aplin, and M. Westwood. 2004. Matrix metalloprotease-3 and -9 proteolyse insulin-like growth factor binding protein-1. Biol. Reprod. 71:438–443. Coussens, L. M., B. Fingleton, and L. M. Matrisian. 2002. Matrix metalloproteinase inhibitors and cancer: trials and tribulations. Science 295: 2387–2392. Cunningham, A. C., K. A. Hasty, J. J. Enghild, and A. E. Mast. 2002. Structural and functional characterization of tissue factor pathway inhibitor following degradation by matrix metalloproteinase-8. Biochem. J. 367:451– 458. Dallas, S. L., J. L. Rosser, G. R. Mundy, and L. F. Bonewald. 2002. Proteolysis of latent transforming growth factor-beta (TGF-beta)-binding protein-1 by osteoclasts. A cellular mechanism for release of TGF-beta from bone matrix. J. Biol. Chem. 277:21352–21360. Dean, R. A., G. S. Butler, Y. Hamma-Kourbali, J. Delbe, D. R. Brigstock, J. Courty, and C. M. Overall. 2007. Identification of candidate angiogenic inhibitors processed by matrix metalloproteinase 2 (MMP-2) in cell-based proteomic screens: disruption of vascular endothelial growth factor (VEGF)/heparin affin regulatory peptide (pleiotrophin) and VEGF/connective tissue growth factor angiogenic inhibitory complexes by MMP-2 proteolysis. Mol. Cell. Biol. 27:8454–8465. Dean, R. A., and C. M. Overall. 2007. Proteomics discovery of metalloproteinase substrates in the cellular context by iTRAQ labeling reveals a diverse MMP-2 substrate degradome. Mol. Cell. Proteomics 6:611–623. De Ceuninck, F., F. Allain, A. Caliez, G. Spik, and P. M. Vanhoutte. 2003. High binding capacity of cyclophilin B to chondrocyte heparan sulfate proteoglycans and its release from the cell surface by matrix metalloproteinases: possible role as a proinflammatory mediator in arthritis. Arthritis Rheum. 48:2197–2206. Diarra, D., M. Stolina, K. Polzer, J. Zwerina, M. S. Ominsky, D. Dwyer, A. Korb, J. Smolen, M. Hoffmann, C. Scheinecker, D. van der Heide, R. Landewe, D. Lacey, W. G. Richards, and G. Schett. 2007. Dickkopf-1 is a master regulator of joint remodeling. Nat. Med. 13:156–163. d’Ortho, M. P., H. Will, S. Atkinson, G. Butler, A. Messent, J. Gavrilovic, B. Smith, R. Timpl, L. Zardi, and G. Murphy. 1997. Membrane-type matrix metalloproteinases 1 and 2 exhibit broad-spectrum proteolytic capacities comparable to many matrix metalloproteinases. Eur. J. Biochem. 250:751– 757. Drummond, A. H., P. Beckett, P. D. Brown, E. A. Bone, A. H. Davidson, W. A. Galloway, A. J. Gearing, P. Huxley, D. Laber, M. McCourt, M. Whittaker, L. M. Wood, and A. Wright. 1999. Preclinical and clinical studies of MMP inhibitors in cancer. Ann. N. Y. Acad. Sci. 878:228–235. Egeblad, M., and Z. Werb. 2002. New functions for the matrix metalloproteinases in cancer progression. Nat. Rev. Cancer 2:161–174. English, W. R., X. S. Puente, J. M. Freije, V. Knauper, A. Amour, A. Merryweather, C. Lopez-Otin, and G. Murphy. 2000. Membrane type 4 matrix metalloproteinase (MMP17) has tumor necrosis factor-alpha convertase activity but does not activate pro-MMP2. J. Biol. Chem. 275:14046– 14055.

4912

BUTLER ET AL.

33. English, W. R., G. Velasco, J. O. Stracke, V. Knauper, and G. Murphy. 2001. Catalytic activities of membrane-type 6 matrix metalloproteinase (MMP25). FEBS Lett. 491:137–142. 34. Eustace, B. K., and D. G. Jay. 2004. Extracellular roles for the molecular chaperone, hsp90. Cell Cycle 3:1098–1100. 35. Fernandez-Patron, C., M. W. Radomski, and S. T. Davidge. 1999. Vascular matrix metalloproteinase-2 cleaves big endothelin-1 yielding a novel vasoconstrictor. Circ. Res. 85:906–911. 36. Fowlkes, J. L., J. J. Enghild, K. Suzuki, and H. Nagase. 1994. Matrix metalloproteinases degrade insulin-like growth factor-binding protein-3 in dermal fibroblast cultures. J. Biol. Chem. 269:25742–25746. 37. Fukai, F., M. Ohtaki, N. Fujii, H. Yajima, T. Ishii, Y. Nishizawa, K. Miyazaki, and T. Katayama. 1995. Release of biological activities from quiescent fibronectin by a conformational change and limited proteolysis by matrix metalloproteinases. Biochemistry 34:11453–11459. 38. Garlanda, C., B. Bottazzi, A. Bastone, and A. Mantovani. 2005. Pentraxins at the crossroads between innate immunity, inflammation, matrix deposition, and female fertility. Annu. Rev. Immunol. 23:337–366. 39. Giannelli, G., J. Falk-Marzillier, O. Schiraldi, W. G. Stetler-Stevenson, and V. Quaranta. 1997. Induction of cell migration by matrix metalloprotease-2 cleavage of laminin-5. Science 277:225–228. 40. Gomis-Ruth, F. X. 2003. Structural aspects of the metzincin clan of metalloendoproteinases. Mol. Biotechnol. 24:157–202. 41. Greenlee, K. J., Z. Werb, and F. Kheradmand. 2007. Matrix metalloproteinases in lung: multiple, multifarious, and multifaceted. Physiol. Rev. 87:69–98. 42. Gygi, S. P., B. Rist, S. A. Gerber, F. Turecek, M. H. Gelb, and R. Aebersold. 1999. Quantitative analysis of complex protein mixtures using isotopecoded affinity tags. Nat. Biotechnol. 17:994–999. 43. Hashimoto, G., I. Inoki, Y. Fujii, T. Aoki, E. Ikeda, and Y. Okada. 2002. Matrix metalloproteinases cleave connective tissue growth factor and reactivate angiogenic activity of vascular endothelial growth factor 165. J. Biol. Chem. 277:36288–36295. 44. Hattori, T., K. Ohsawa, Y. Mizuno, K. Kato, and S. Kohsaka. 1994. Synthetic peptide corresponding to 30 amino acids of the C-terminal of neuronspecific enolase promotes survival of neocortical neurons in culture. Biochem. Biophys. Res. Commun. 202:25–30. 45. He, Z., and A. Bateman. 2003. Progranulin (granulin-epithelin precursor, PC-cell-derived growth factor, acrogranin) mediates tissue repair and tumorigenesis. J. Mol. Med. 81:600–612. 46. Heathfield, T. F., P. Onnerfjord, L. Dahlberg, and D. Heinegard. 2004. Cleavage of fibromodulin in cartilage explants involves removal of the N-terminal tyrosine sulfate-rich region by proteolysis at a site that is sensitive to matrix metalloproteinase-13. J. Biol. Chem. 279:6286–6295. 47. Hennipman, A., J. Smits, B. van Oirschot, J. C. van Houwelingen, G. Rijksen, J. P. Neyt, J. A. Van Unnik, and G. E. Staal. 1987. Glycolytic enzymes in breast cancer, benign breast disease and normal breast tissue. Tumour Biol. 8:251–263. 48. Henry, M. T., K. McMahon, C. Costello, M. X. Fitzgerald, and C. M. O’Connor. 2002. Secretory leukocyte proteinase inhibitor and elafin are resistant to degradation by MMP-8. Exp. Lung Res. 28:85–97. 49. Hernandez, J. D., and L. G. Baum. 2002. Ah, sweet mystery of death! Galectins and control of cell fate. Glycobiology 12:127R–136R. 50. Hernandez-Barrantes, S., M. Bernardo, M. Toth, and R. Fridman. 2002. Regulation of membrane type-matrix metalloproteinases. Semin. Cancer Biol. 12:131–138. 51. Higashi, S., and K. Miyazaki. 2003. Novel processing of beta-amyloid precursor protein catalyzed by membrane type 1 matrix metalloproteinase releases a fragment lacking the inhibitor domain against gelatinase A. Biochemistry 42:6514–6526. 52. Hiller, O., A. Lichte, A. Oberpichler, A. Kocourek, and H. Tschesche. 2000. Matrix metalloproteinases collagenase-2, macrophage elastase, collagenase-3, and membrane type 1-matrix metalloproteinase impair clotting by degradation of fibrinogen and factor XII. J. Biol. Chem. 275:33008–33013. 53. Honbou, K., N. N. Suzuki, M. Horiuchi, T. Niki, T. Taira, H. Ariga, and F. Inagaki. 2003. The crystal structure of DJ-1, a protein related to male fertility and Parkinson’s disease. J. Biol. Chem. 278:31380–31384. 54. Hopwood, J. J., S. Bunge, C. P. Morris, P. J. Wilson, C. Steglich, M. Beck, E. Schwinger, and A. Gal. 1993. Molecular basis of mucopolysaccharidosis type II: mutations in the iduronate-2-sulphatase gene. Hum. Mutat. 2:435– 442. 55. Jin, Z. G., A. O. Lungu, L. Xie, M. Wang, C. Wong, and B. C. Berk. 2004. Cyclophilin A is a proinflammatory cytokine that activates endothelial cells. Arterioscler. Thromb. Vasc. Biol. 24:1186–1191. 56. Jurianz, K., S. Ziegler, H. Garcia-Schuler, S. Kraus, O. Bohana-Kashtan, Z. Fishelson, and M. Kirschfink. 1999. Complement resistance of tumor cells: basal and induced mechanisms. Mol. Immunol. 36:929–939. 57. Kaiser, E., R. Kuzmits, P. Pregant, O. Burghuber, and W. Worofka. 1989. Clinical biochemistry of neuron specific enolase. Clin. Chim. Acta 183: 13–31. 58. Kimberley, F. C., B. Sivasankar, and B. Paul Morgan. 2007. Alternative roles for CD59. Mol. Immunol. 44:73–81.

MOL. CELL. BIOL. 59. Knauper, V., H. Will, C. Lopez-Otin, B. Smith, S. J. Atkinson, H. Stanton, R. M. Hembry, and G. Murphy. 1996. Cellular mechanisms for human procollagenase-3 (MMP-13) activation. Evidence that MT1-MMP (MMP14) and gelatinase a (MMP-2) are able to generate active enzyme. J. Biol. Chem. 271:17124–17131. 60. Koolwijk, P., N. Sidenius, E. Peters, C. F. Sier, R. Hanemaaijer, F. Blasi, and V. W. van Hinsbergh. 2001. Proteolysis of the urokinase-type plasminogen activator receptor by metalloproteinase-12: implication for angiogenesis in fibrin matrices. Blood 97:3123–3131. 61. Koshikawa, N., G. Giannelli, V. Cirulli, K. Miyazaki, and V. Quaranta. 2000. Role of cell surface metalloprotease MT1-MMP in epithelial cell migration over laminin-5. J. Cell Biol. 148:615–624. 62. Koshikawa, N., T. Minegishi, A. Sharabi, V. Quaranta, and M. Seiki. 2005. Membrane-type matrix metalloproteinase-1 (MT1-MMP) is a processing enzyme for human laminin gamma 2 chain. J. Biol. Chem. 280:88–93. 63. Kuster, B., M. Schirle, P. Mallick, and R. Aebersold. 2005. Scoring proteomes with proteotypic peptide probes. Nat. Rev. Mol. Cell Biol. 6:577– 583. 64. Lee, S., S. M. Jilani, G. V. Nikolova, D. Carpizo, and M. L. Iruela-Arispe. 2005. Processing of VEGF-A by matrix metalloproteinases regulates bioavailability and vascular patterning in tumors. J. Cell Biol. 169:681–691. 65. Le Naour, F., D. E. Misek, M. C. Krause, L. Deneux, T. J. Giordano, S. Scholl, and S. M. Hanash. 2001. Proteomics-based identification of RS/ DJ-1 as a novel circulating tumor antigen in breast cancer. Clin. Cancer Res. 7:3328–3335. 66. Lev, N., D. Roncevich, D. Ickowicz, E. Melamed, and D. Offen. 2006. Role of DJ-1 in Parkinson’s disease. J. Mol. Neurosci. 29:215–225. 67. Li, Q., P. W. Park, C. L. Wilson, and W. C. Parks. 2002. Matrilysin shedding of syndecan-1 regulates chemokine mobilization and transepithelial efflux of neutrophils in acute lung injury. Cell 111:635–646. 68. Liang, S. L., and D. W. Chan. 2007. Enzymes and related proteins as cancer biomarkers: a proteomic approach. Clin. Chim. Acta 381:93–97. 69. Lindsay, M. A. 2003. Target discovery. Nat. Rev. Drug Discov. 2:831–838. 70. Liou, H. L., S. S. Dixit, S. Xu, G. S. Tint, A. M. Stock, and P. Lobel. 2006. NPC2, the protein deficient in Niemann-Pick C2 disease, consists of multiple glycoforms that bind a variety of sterols. J. Biol. Chem. 281:36710– 36723. 71. Lohi, J., K. Lehti, J. Westermarck, V. M. Kahari, and J. Keski-Oja. 1996. Regulation of membrane-type matrix metalloproteinase-1 expression by growth factors and phorbol 12-myristate 13-acetate. Eur. J. Biochem. 239: 239–247. 72. Lope´z-Otı´n, C., and C. M. Overall. 2002. Protease degradomics: a new challenge for proteomics. Nat. Rev. Mol. Cell Biol. 3:509–519. 73. Manes, S., M. Llorente, R. A. Lacalle, C. Gomez-Mouton, L. Kremer, E. Mira, and A. C. Martinez. 1999. The matrix metalloproteinase-9 regulates the insulin-like growth factor-triggered autocrine response in DU-145 carcinoma cells. J. Biol. Chem. 274:6935–6945. 74. Manes, S., E. Mira, M. M. Barbacid, A. Cipres, P. Fernandez-Resa, J. M. Buesa, I. Merida, M. Aracil, G. Marquez, and A. C. Martinez. 1997. Identification of insulin-like growth factor-binding protein-1 as a potential physiological substrate for human stromelysin-3. J. Biol. Chem. 272:25706– 25712. 75. McCawley, L. J., and L. M. Matrisian. 2001. Matrix metalloproteinases: they’re not just for matrix anymore! Curr. Opin. Cell Biol. 13:534–540. 76. McQuibban, G. A., J. H. Gong, E. M. Tam, C. A. McCulloch, I. ClarkLewis, and C. M. Overall. 2000. Inflammation dampened by gelatinase A cleavage of monocyte chemoattractant protein-3. Science 289:1202–1206. 77. McQuibban, G. A., J. H. Gong, J. P. Wong, J. L. Wallace, I. Clark-Lewis, and C. M. Overall. 2002. Matrix metalloproteinase processing of monocyte chemoattractant proteins generates CC chemokine receptor antagonists with anti-inflammatory properties in vivo. Blood 100:1160–1167. 78. Melaragno, M. G., Y. W. Fridell, and B. C. Berk. 1999. The Gas6/Axl system: a novel regulator of vascular cell function. Trends Cardiovasc. Med. 9:250–253. 79. Miyamoto, S., K. Yano, S. Sugimoto, G. Ishii, T. Hasebe, Y. Endoh, K. Kodama, M. Goya, T. Chiba, and A. Ochiai. 2004. Matrix metalloproteinase-7 facilitates insulin-like growth factor bioavailability through its proteinase activity on insulin-like growth factor binding protein 3. Cancer Res. 64:665–671. 80. Miyazaki, K., M. Hasegawa, K. Funahashi, and M. Umeda. 1993. A metalloproteinase inhibitor domain in Alzheimer amyloid protein precursor. Nature 362:839–841. 81. Myint, E., D. J. Brown, A. V. Ljubimov, M. Kyaw, and M. C. Kenney. 1996. Cleavage of human corneal type VI collagen alpha 3 chain by matrix metalloproteinase-2. Cornea 15:490–496. 82. Nagakubo, D., T. Taira, H. Kitaura, M. Ikeda, K. Tamai, S. M. IguchiAriga, and H. Ariga. 1997. DJ-1, a novel oncogene which transforms mouse NIH3T3 cells in cooperation with ras. Biochem. Biophys. Res. Commun. 231:509–513. 83. Nuttall, R. K., C. J. Pennington, J. Taplin, A. Wheal, V. W. Yong, P. A. Forsyth, and D. R. Edwards. 2003. Elevated membrane-type matrix metal-

VOL. 28, 2008

84.

85.

86.

87.

88. 89.

90. 91.

92.

93. 94.

95.

96.

97.

98.

99. 100.

101.

102.

103. 104.

105.

106.

107.

PHARMACOPROTEOMICS OF A METALLOPROTEINASE INHIBITOR

loproteinases in gliomas revealed by profiling proteases and inhibitors in human cancer cells. Mol. Cancer Res. 1:333–345. Ochieng, J., R. Fridman, P. Nangia-Makker, D. E. Kleiner, L. A. Liotta, W. G. Stetler-Stevenson, and A. Raz. 1994. Galectin-3 is a novel substrate for human matrix metalloproteinases-2 and -9. Biochemistry 33:14109– 14114. Ochieng, J., B. Green, S. Evans, O. James, and P. Warfield. 1998. Modulation of the biological functions of galectin-3 by matrix metalloproteinases. Biochim. Biophys. Acta 1379:97–106. Olzmann, J. A., K. Brown, K. D. Wilkinson, H. D. Rees, Q. Huai, H. Ke, A. I. Levey, L. Li, and L. S. Chin. 2004. Familial Parkinson’s diseaseassociated L166P mutation disrupts DJ-1 protein folding and function. J. Biol. Chem. 279:8506–8515. O’Reilly, M. S., L. Holmgren, Y. Shing, C. Chen, R. A. Rosenthal, M. Moses, W. S. Lane, Y. Cao, E. H. Sage, and J. Folkman. 1994. Angiostatin: a novel angiogenesis inhibitor that mediates the suppression of metastases by a Lewis lung carcinoma. Cell 79:315–328. Overall, C. M. 2004. Dilating the degradome: matrix metalloproteinase 2 (MMP-2) cuts to the heart of the matter. Biochem. J. 383:e5–7. Overall, C. M. 2002. Molecular determinants of metalloproteinase substrate specificity: matrix metalloproteinase substrate binding domains, modules, and exosites. Mol. Biotechnol. 22:51–86. Overall, C. M., and C. P. Blobel. 2007. In search of partners: linking extracellular proteases to substrates. Nat. Rev. Mol. Cell Biol. 8:245–257. Overall, C. M., and R. A. Dean. 2006. Degradomics: systems biology of the protease web. Pleiotropic roles of MMPs in cancer. Cancer Metastasis Rev. 25:69–75. Overall, C. M., and O. Kleifeld. 2006. Tumour microenvironment–opinion: validating matrix metalloproteinases as drug targets and anti-targets for cancer therapy. Nat. Rev. Cancer 6:227–239. Overall, C. M., and C. Lope´z-Otı´n. 2002. Strategies for MMP inhibition in cancer: innovations for the post-trial era. Nat. Rev. Cancer 2:657–672. Overall, C. M., G. A. McQuibban, and I. Clark-Lewis. 2002. Discovery of chemokine substrates for matrix metalloproteinases by exosite scanning: a new tool for degradomics. Biol. Chem. 383:1059–1066. Page-McCaw, A., A. J. Ewald, and Z. Werb. 2007. Matrix metalloproteinases and the regulation of tissue remodelling. Nat. Rev. Mol. Cell Biol. 8:221–233. Pardo, M., A. Garcia, B. Thomas, A. Pineiro, A. Akoulitchev, R. A. Dwek, and N. Zitzmann. 2006. The characterization of the invasion phenotype of uveal melanoma tumour cells shows the presence of MUC18 and HMG-1 metastasis markers and leads to the identification of DJ-1 as a potential serum biomarker. Int. J. Cancer 119:1014–1022. Parks, W. C., C. L. Wilson, and Y. S. Lopez-Boado. 2004. Matrix metalloproteinases as modulators of inflammation and innate immunity. Nat. Rev. Immunol. 4:617–629. Parnell, P. G., J. Wunderlich, B. Carter, and J. Halper. 1992. Transforming growth factor e: amino acid analysis and partial amino acid sequence. Growth Factors 7:65–72. Perbal, B. 2004. CCN proteins: multifunctional signalling regulators. Lancet 363:62–64. Pike, S. E., L. Yao, K. D. Jones, B. Cherney, E. Appella, K. Sakaguchi, H. Nakhasi, J. Teruya-Feldstein, P. Wirth, G. Gupta, and G. Tosato. 1998. Vasostatin, a calreticulin fragment, inhibits angiogenesis and suppresses tumor growth. J. Exp. Med. 188:2349–2356. Pirila, E., A. Sharabi, T. Salo, V. Quaranta, H. Tu, R. Heljasvaara, N. Koshikawa, T. Sorsa, and P. Maisi. 2003. Matrix metalloproteinases process the laminin-5 gamma 2-chain and regulate epithelial cell migration. Biochem. Biophys. Res. Commun. 303:1012–1017. Puente, X. S., L. M. Sanchez, C. M. Overall, and C. Lopez-Otin. 2003. Human and mouse proteases: a comparative genomic approach. Nat. Rev. Genet. 4:544–558. Rabinovich, G. A. 2005. Galectin-1 as a potential cancer target. Br. J. Cancer 92:1188–1192. Rath, G. M., C. Schneider, S. Dedieu, B. Rothhut, M. Soula-Rothhut, C. Ghoneim, B. Sid, H. Morjani, H. El Btaouri, and L. Martiny. 2006. The C-terminal CD47/IAP-binding domain of thrombospondin-1 prevents camptothecin- and doxorubicin-induced apoptosis in human thyroid carcinoma cells. Biochim. Biophys. Acta 1763:1125–1134. Restifo, N. P., F. M. Marincola, Y. Kawakami, J. Taubenberger, J. R. Yannelli, and S. A. Rosenberg. 1996. Loss of functional beta 2-microglobulin in metastatic melanomas from five patients receiving immunotherapy. J. Natl. Cancer Inst. 88:100–108. Rofstad, E. K., K. Henriksen, K. Galappathi, and B. Mathiesen. 2003. Antiangiogenic treatment with thrombospondin-1 enhances primary tumor radiation response and prevents growth of dormant pulmonary micrometastases after curative radiation therapy in human melanoma xenografts. Cancer Res. 63:4055–4061. Rozanov, D. V., B. Ghebrehiwet, T. I. Postnova, A. Eichinger, E. I. Deryugina, and A. Y. Strongin. 2002. The hemopexin-like C-terminal domain of membrane type 1 matrix metalloproteinase regulates proteolysis of a multifunctional protein, gC1qR. J. Biol. Chem. 277:9318–9325.

4913

108. Rozanov, D. V., A. Y. Savinov, V. S. Golubkov, T. I. Postnova, A. Remacle, S. Tomlinson, and A. Y. Strongin. 2004. Cellular membrane type-1 matrix metalloproteinase (MT1-MMP) cleaves C3b, an essential component of the complement system. J. Biol. Chem. 279:46551–46557. 109. Rubinstein, N., J. M. Ilarregui, M. A. Toscano, and G. A. Rabinovich. 2004. The role of galectins in the initiation, amplification and resolution of the inflammatory response. Tissue Antigens 64:1–12. 110. Rusk, A., E. McKeegan, F. Haviv, S. Majest, J. Henkin, and C. Khanna. 2006. Preclinical evaluation of antiangiogenic thrombospondin-1 peptide mimetics, ABT-526 and ABT-510, in companion dogs with naturally occurring cancers. Clin. Cancer Res. 12:7444–7455. 111. Sadowski, T., S. Dietrich, F. Koschinsky, and R. Sedlacek. 2003. Matrix metalloproteinase 19 regulates insulin-like growth factor-mediated proliferation, migration, and adhesion in human keratinocytes through proteolysis of insulin-like growth factor binding protein-3. Mol. Biol. Cell 14:4569– 4580. 112. Sato, H., T. Takino, Y. Okada, J. Cao, A. Shinagawa, E. Yamamoto, and M. Seiki. 1994. A matrix metalloproteinase expressed on the surface of invasive tumour cells. Nature 370:61–65. 113. Saunders, W. B., K. J. Bayless, and G. E. Davis. 2005. MMP-1 activation by serine proteases and MMP-10 induces human capillary tubular network collapse and regression in 3D collagen matrices. J. Cell Sci. 118:2325–2340. 114. Schalkwijk, J., O. Wiedow, and S. Hirose. 1999. The trappin gene family: proteins defined by an N-terminal transglutaminase substrate domain and a C-terminal four-disulphide core. Biochem. J. 340:569–577. 115. Selakovic, V., R. Raicevic, and L. Radenovic. 2005. The increase of neuronspecific enolase in cerebrospinal fluid and plasma as a marker of neuronal damage in patients with acute brain infarction. J. Clin. Neurosci. 12:542– 547. 116. Shalinsky, D. R., J. Brekken, H. Zou, L. A. Bloom, C. D. McDermott, S. Zook, N. M. Varki, and K. Appelt. 1999. Marked antiangiogenic and antitumor efficacy of AG3340 in chemoresistant human non-small cell lung cancer tumors: single agent and combination chemotherapy studies. Clin. Cancer Res. 5:1905–1917. 117. Shalinsky, D. R., J. Brekken, H. Zou, C. D. McDermott, P. Forsyth, D. Edwards, S. Margosiak, S. Bender, G. Truitt, A. Wood, N. M. Varki, and K. Appelt. 1999. Broad antitumor and antiangiogenic activities of AG3340, a potent and selective MMP inhibitor undergoing advanced oncology clinical trials. Ann. N. Y. Acad. Sci. 878:236–270. 118. Sid, B., H. Sartelet, G. Bellon, H. El Btaouri, G. Rath, N. Delorme, B. Haye, and L. Martiny. 2004. Thrombospondin 1: a multifunctional protein implicated in the regulation of tumor growth. Crit. Rev. Oncol. Hematol. 49: 245–258. 119. Sidis, Y., A. Mukherjee, H. Keutmann, A. Delbaere, M. Sadatsuki, and A. Schneyer. 2006. Biological activity of follistatin isoforms and follistatinlike-3 is dependent on differential cell surface binding and specificity for activin, myostatin, and bone morphogenetic proteins. Endocrinology 147: 3586–3597. 120. Sole, S., V. Petegnief, R. Gorina, A. Chamorro, and A. M. Planas. 2004. Activation of matrix metalloproteinase-3 and agrin cleavage in cerebral ischemia/reperfusion. J. Neuropathol. Exp. Neurol. 63:338–349. 121. Sternlicht, M. D., and Z. Werb. 1999. ECM proteinases, p. 503–563. In T. Kreis and R. Vale (ed.), Guidebook to the extracellular matrix, anchor and adhesion proteins. Oxford University Press, Oxford, United Kingdom. 122. Sternlicht, M. D., and Z. Werb. 2001. How matrix metalloproteinases regulate cell behavior. Annu. Rev. Cell Dev. Biol. 17:463–516. 123. Stix, B., T. Kahne, K. Sletten, J. Raynes, A. Roessner, and C. Rocken. 2001. Proteolysis of AA amyloid fibril proteins by matrix metalloproteinases-1, -2, and -3. Am. J. Pathol. 159:561–570. 124. Stracke, J. O., M. Hutton, M. Stewart, A. M. Pendas, B. Smith, C. LopezOtin, G. Murphy, and V. Knauper. 2000. Biochemical characterization of the catalytic domain of human matrix metalloproteinase 19. Evidence for a role as a potent basement membrane degrading enzyme. J. Biol. Chem. 275:14809–14816. 125. Strongin, A. Y., I. Collier, G. Bannikov, B. L. Marmer, G. A. Grant, and G. I. Goldberg. 1995. Mechanism of cell surface activation of 72-kDa type IV collagenase. Isolation of the activated form of the membrane metalloprotease. J. Biol. Chem. 270:5331–5338. 126. Sugita, Y., and Y. Masuho. 1995. CD59: its role in complement regulation and potential for therapeutic use. Immunotechnology 1:157–168. 127. Takino, T., N. Koshikawa, H. Miyamori, M. Tanaka, T. Sasaki, Y. Okada, M. Seiki, and H. Sato. 2003. Cleavage of metastasis suppressor gene product KiSS-1 protein/metastin by matrix metalloproteinases. Oncogene 22: 4617–4626. 128. Tam, E. M., T. R. Moore, G. S. Butler, and C. M. Overall. 2004. Characterization of the distinct collagen binding, helicase and cleavage mechanisms of matrix metalloproteinase 2 and 14 (gelatinase A and MT1-MMP): the differential roles of the MMP hemopexin C domains and the MMP-2 fibronectin type II modules in collagen triple helicase activities. J. Biol. Chem. 279:43336–43344. 129. Tam, E. M., C. J. Morrison, Y. I. Wu, M. S. Stack, and C. M. Overall. 2004. Membrane protease proteomics: isotope-coded affinity tag MS identifica-

4914

130.

131.

132.

133.

134.

135.

136.

137.

138. 139.

BUTLER ET AL.

tion of undescribed MT1-matrix metalloproteinase substrates. Proc. Natl. Acad. Sci. USA 101:6917–6922. Tam, E. M., Y. I. Wu, G. S. Butler, M. S. Stack, and C. M. Overall. 2002. Collagen binding properties of the membrane type-1 matrix metalloproteinase (MT1-MMP) hemopexin C domain. The ectodomain of the 44-kDa autocatalytic product of MT1-MMP inhibits cell invasion by disrupting native type I collagen cleavage. J. Biol. Chem. 277:39005–39014. Tester, A. M., J. H. Cox, A. R. Connor, A. E. Starr, R. A. Dean, X. S. Puente, C. Lopez-Otin, and C. M. Overall. 2007. LPS responsiveness and neutrophil chemotaxis in vivo require PMN MMP-8 activity. PLoS ONE 2:e312. Tierney, G. M., N. R. Griffin, R. C. Stuart, H. Kasem, K. P. Lynch, J. T. Lury, P. D. Brown, A. W. Millar, R. J. Steele, and S. L. Parsons. 1999. A pilot study of the safety and effects of the matrix metalloproteinase inhibitor marimastat in gastric cancer. Eur. J. Cancer 35:563–568. Toth, M., S. Hernandez-Barrantes, P. Osenkowski, M. M. Bernardo, D. C. Gervasi, Y. Shimura, O. Meroueh, L. P. Kotra, B. G. Galvez, A. G. Arroyo, S. Mobashery, and R. Fridman. 2002. Complex pattern of membrane type 1 matrix metalloproteinase shedding. Regulation by autocatalytic cells surface inactivation of active enzyme. J. Biol. Chem. 277:26340–26350. Toth, M., P. Osenkowski, D. Hesek, S. Brown, S. Meroueh, W. Sakr, S. Mobashery, and R. Fridman. 2005. Cleavage at the stem region releases an active ectodomain of the membrane type 1 matrix metalloproteinase. Biochem. J. 387:497–506. Ugwu, F., B. Van Hoef, A. Bini, D. Collen, and H. R. Lijnen. 1998. Proteolytic cleavage of urokinase-type plasminogen activator by stromelysin-1 (MMP-3). Biochemistry 37:7231–7236. van Beijnum, J. R., R. P. Dings, E. van der Linden, B. M. Zwaans, F. C. Ramaekers, K. H. Mayo, and A. W. Griffioen. 2006. Gene expression of tumor angiogenesis dissected: specific targeting of colon cancer angiogenic vasculature. Blood 108:2339–2348. Van den Steen, P. E., P. Proost, A. Wuyts, J. Van Damme, and G. Opdenakker. 2000. Neutrophil gelatinase B potentiates interleukin-8 tenfold by aminoterminal processing, whereas it degrades CTAP-III, PF-4, and GROalpha and leaves RANTES and MCP-2 intact. Blood 96:2673–2681. VanSaun, M., and M. J. Werle. 2000. Matrix metalloproteinase-3 removes agrin from synaptic basal lamina. J. Neurobiol. 43:140–149. Waragai, M., J. Wei, M. Fujita, M. Nakai, G. J. Ho, E. Masliah, H. Akatsu, T. Yamada, and M. Hashimoto. 2006. Increased level of DJ-1 in the cerebrospinal fluids of sporadic Parkinson’s disease. Biochem. Biophys. Res. Commun. 345:967–972.

MOL. CELL. BIOL. 140. Wilkinson, L., G. Kolle, D. Wen, M. Piper, J. Scott, and M. Little. 2003. CRIM1 regulates the rate of processing and delivery of bone morphogenetic proteins to the cell surface. J. Biol. Chem. 278:34181–34188. 141. Will, H., S. J. Atkinson, G. S. Butler, B. Smith, and G. Murphy. 1996. The soluble catalytic domain of membrane type 1 matrix metalloproteinase cleaves the propeptide of progelatinase A and initiates autoproteolytic activation. Regulation by TIMP-2 and TIMP-3. J. Biol. Chem. 271:17119– 17123. 142. Wu, H. B., C. Y. Lee, and M. M. Rechler. 1999. Proteolysis of insulin-like growth factor binding protein-3 in serum from pregnant, non-pregnant and fetal rats by matrix metalloproteinases and serine proteases. Horm. Metab. Res. 31:186–191. 143. Yamada, H., F. Saito, H. Fukuta-Ohi, D. Zhong, A. Hase, K. Arai, A. Okuyama, R. Maekawa, T. Shimizu, and K. Matsumura. 2001. Processing of beta-dystroglycan by matrix metalloproteinase disrupts the link between the extracellular matrix and cell membrane via the dystroglycan complex. Hum. Mol. Genet. 10:1563–1569. 144. Yurchenko, V., G. Zybarth, M. O’Connor, W. W. Dai, G. Franchin, T. Hao, H. Guo, H. C. Hung, B. Toole, P. Gallay, B. Sherry, and M. Bukrinsky. 2002. Active site residues of cyclophilin A are crucial for its signaling activity via CD147. J. Biol. Chem. 277:22959–22965. 145. Zhang, K., G. A. McQuibban, C. Silva, G. S. Butler, J. B. Johnston, J. Holden, I. Clark-Lewis, C. M. Overall, and C. Power. 2003. HIV-induced metalloproteinase processing of the chemokine stromal cell derived factor-1 causes neurodegeneration. Nat. Neurosci. 6:1064–1071. 146. Zhang, X., C. Connolly, M. Duquette, J. Lawler, and S. Parangi. 2007. Continuous administration of the three thrombospondin-1 type 1 repeats recombinant protein improves the potency of therapy in an orthotopic human pancreatic cancer model. Cancer Lett. 247:143–149. 147. Zhang, X., J. Xu, J. Lawler, E. Terwilliger, and S. Parangi. 2007. Adenoassociated virus-mediated antiangiogenic gene therapy with thrombospondin-1 type 1 repeats and endostatin. Clin. Cancer Res. 13:3968– 3976. 148. Zhong, D., F. Saito, Y. Saito, A. Nakamura, T. Shimizu, and K. Matsumura. 2006. Characterization of the protease activity that cleaves the extracellular domain of beta-dystroglycan. Biochem. Biophys. Res. Commun. 345:867– 871. 149. Zucker, S., J. Cao, and W. T. Chen. 2000. Critical appraisal of the use of matrix metalloproteinase inhibitors in cancer treatment. Oncogene 19:6642–6650.